WWDC2004 Session 605

Transcript

Kind: captions
Language: en
oh and welcome thank you for joining us
today I think you're going to find it's
a very useful session especially in your
market where you need to do backups my
name is Chris bloodstone the senior
Enterprise Alliance manager and I manage
all of our backup vendors here for Apple
computer so let's get going have you
ever had this happen to you back in 1996
I work for Oracle and we had this proc
that was called oracle card does anyone
remember oracle card all right well
anyway I was building this presentation
on this mac so I could go to a show
similar to this and as I went to the
show late night you know you're doing
your coding work or getting your demo
set up and boom my hard drive crashed
like three in the morning on a Thursday
night you like great so literally for
the next eight hours it had to recreate
all the material that i had and sure
enough i didn't have a backup so for me
as an individual for doing a
presentation eight hours was difficult
or not but an enterprise it's absolutely
critical that kind of time for an
enterprise can easily represent millions
of dollars of cost to you and your
enterprising your business that's why
it's absolutely critical that you have
backup so some of things going to talk
about today some of the backup trends
that we're seeing occurring in the
market today everything from regulation
compliance to the types of backup
windows that we run into also what I'd
like to talk about is our backup market
segments what we do is we kind of look
at the whole market and said okay what
type of backup needs to our customers
for Apple really need and I'll talking
more detail about whether it's an
individual system a laptop a powermac a
powerbook all the way scaling up to
large sand type deployments like that
our xn product line so the goal of this
session is so that when you get done
you'll have an idea of okay what kind of
backup solution can I use because
there's a number of vendors out there
available for you as well as find out
what solutions work on the platform
today so before I dig into all of the
backup materials like to cover a few
terms first the first one is fibre
channel fibre channel is a technology
for transmitting data from your computer
service
across your network so it's a
communication capability it's really
great for fibre channel when you have
like storage devices that are hooked up
to the network which kind of leads into
the next idea which is called raid
redundant array of inexpensive disks or
sometimes they're called independent
disk because it says on the screen the
idea with the rate is that instead of
having a single disc which has all your
data and it could fail you set up a ray
or a number of those disks and typically
you want to do that for a couple reasons
first of all you want to do it so you
can get performance so instead of
reading from a single disk you can read
across a set of discs and be able to
access that information very quickly
also you want to do it for our fault
tolerance so in case the drift goes down
you either have replications of that
data on the disk so you can get access
to it we're not going to really drill
down into the raid today like the root
different raid levels we have a great
session on xserve deployment and for
radix sort of raid deployment that you
should definitely be checking out and we
talked a great deal about that also
there's this concept of network attached
storage right so what you do here
instead of taking your hard drive
storage like a raid server and attacking
it to your departmental server what you
do is you break it away and give it its
own independent network address the
advantage then is that your applications
that are running on the server don't
have to compete on the processor with
your disk access so by separating and
putting your your data on the network it
makes it much more efficient for your
server to get access to the information
and still have access to the stat in a
very fast and efficient way which leads
onto so you've got this network attached
to storage area networks this is a
network that oftentimes based on fibre
channel they can be some other
technologies where you have multiple
disks that are on your network talking
to multiple servers that are also on
this high-speed interconnect technology
which so that's what a fan is now that
leads to what's called HSM hierarchical
storage management what that does is it
says ok now I've got these discs on my
network how do I manage the migration of
that data from one
set to the other and that's what HSM
tries to address is okay how do I manage
it which leads to the last term ilm
information lifecycle management and
that's managing the whole stack from
birth of the data when it's created to
its obsolescence when you no longer need
it you could actually safely delete it
and move on so those just want to make
sure you understand some of the
terminology that we'll be covering today
in our presentation storage requirements
anyone here have stock in a world com
okay because of those types of scenarios
that happen in our market we have the
sarbanes-oxley Act and what that act
says is that for four years you must
keep all of your audit records so if
you're a corporation you need to keep
those records for four years well then
it gives you in more challenging in the
pharmaceutical life sciences let's say
for example you're a pharmaceutical
company that does viagra well the
company that runs that product has to
keep the information for that product
for an additional five years after that
product might have been off celeste and
moved on so they have to keep track of
that for example financial services if
you're a broker in the financial
services organization they're required
to hold onto the information to the
stuff that you've been doing for
literally the life cycle of the entire
company so if you're a Charl swath for
example you have to keep track of all
that information as long as you exist as
a company moving on the health care a
lot of the health care regulatory
requirements is that now you have to
keep records not only till you're 21
years old like if you're born in a
hospital they tracker for 21 years but
even like five or six years after you
passed away and gone on they have to
keep track of that information and who
can forget OSHA so if you work for OSHA
let's say you're working with toxic
substances theylor they have to keep
that information for 30 years even after
you moved out of OSHA so you can see
that these types of requirement is
creating an explosion in what we need
for storage capability companies like
IDC and stuff and forecasted it's at
around sixty four percent compounded
growth based just on regulatory
compliance needs so as you can see from
a petabyte perspective that's morn
more data that you have to keep track of
in years the compounded you've got these
backup windows that are getting smaller
and smaller so what you see here in this
chart is okay based upon the type of
tech to take technology i'm using it
takes a long time to back up a terabyte
so even if you're doing like disc the
disc and i pulled some of my good back
up vendors here you're typically looking
anywhere from two to four hours
depending on what kind of throughput you
can actually push through your fibre
channel network so that's just the back
of a terabyte so you can imagine as
you're working in high video demand type
environments the need to be able to
shrink that back i'm going to be able to
dress it effectively so you can still
get access to the data you need and
safely archive it so now what I'd like
to talk about it's what kind of
architecture so when you're thinking of
backups there's a number of ways you can
actually create it I picked for the ones
that are probably the most recognized
and most widely use in enterprises today
the first one is just a regular backup
server very very straightforward what
you see here is you gotta x server right
X an xserve g5 talking to the fibre
channel switch connected to X serve read
into a tape library so very easy to back
it up goes through the server takes it
from the tape from the hard drive pops
it in the tape very easy so this
actually makes a lot of sense right you
kind of got this one-to-one relationship
so people can kind of wrap their heads
around it and get that get the fast back
up because you're going directly from
one medium to another it supports these
automatic tape libraries which everyone
really has to have to be able to get
access to your data and really have a
relatively fast recovery because you
have this nice bandwidth it's dedicated
to getting access to your data and
backing it up however there's some
problems with this type of setup for
example your backup resources it's not
really really good use right because
you're not always using the tape library
and there's actually other methods that
allow you to get better use of those
resources and honestly it just doesn't
scale you can't afford for every you
know three or four rate boxes be able to
have a single tape library it just
doesn't scale well and the other thing
that you've seen the half of
exceptionally large enterprises is that
each administrator for that server
becomes their back
person so you end up having a lot of
administrators backing up their own
servers so if those administrators are
heterogeneous in nature you know Apple
Linux etc then what they end up doing is
they have to be specialized on their own
specific server which kind of led up to
this idea of backup client this is
probably the most popular and most
well-known so the idea here is that you
take your backup client and in a backup
server the server talks for the client
says okay today is Wednesday at noon
trying to take all your data and put it
on the tape library so it's a nice very
easy disk to tape type of backup
scenario this is great if you have a ton
of different operating systems in your
network all of my vendors here have
client for not only on our platform but
also across all the different
permutations I've talked to some that
have it up to 60 different operating
systems where they have client back up
so that they can actually get access to
their clients and back it up to the
server so generally you have good
performance right depends that depends
on how fast your actual network is that
your backup server is connected to your
backup client that really defines it and
it's very easy to add new platforms
right because all you got to do you call
up your vendor you go hey I added XYZ
box I need to get a client for that they
put it on there and it works within your
existing infrastructure that's why it's
so popular but there's some problem
right for example if you have a large
backup and restore or a large back with
window like we saw on that earlier slide
for backing up just a terabyte it's a
little hard to get it timed up right and
then typically what you want to do is
this thing called multiplexing where
you're writing the multiple tape
libraries at the same time because you
get you're trying to backup many people
in a similar time frame like overnight
and it really doesn't solve the problem
these large backup lenders if you're
backing up let's say five terabytes and
I kind of hard to do that with just tape
backup which led to the evolution of
what's called disc the disc the tape
which probably a lotta you heard about
very straightforward concept so what you
do is you take your backup client and
you back it directly to disk very quick
very efficient using a maximum bandwidth
there and usually do that a more
periodic basis
and then as your data gets old you know
like an ilm type of technology or gone
okay I no longer need to keep this on my
desk because you know I've had it backed
up for a month I'm probably not going to
look at that data when I just go ahead
and archive it and archiving is
definitely very important because when
you archive it that's when you usually
take it off site right so that way if
there's a fire or there's some other
kind of physical danger thing that
happens to your data you've got off-site
storage you can get access to it so
that's typically would you do this disk
to disk the tape what's really really
good about this is you get this
centralized administration all right you
don't have the problem where you have a
bunch of different administrators
managing a bunch of different servers
you generally can reduce the number
administrators you have focus
exclusively on backup it also gives you
easy access to this critical data so for
example if you're doing a back up every
day or every hour or every half a day
right you can get access to the data
that you know you actually do in the
file I need to go get that information
and I got to put it back on you know
you're able to do that pretty easily it
also takes advantage of the great throop
but you can get to your dish alright so
now you're getting just the dish you're
taking advantage you're not being slowed
down by doing your archival back up and
then you migrate the data right as it
gets less expensive media so instead of
singing it on your drive you can
actually put it up onto your tape
there's some challenges obviously with
that right so the more moving parts you
have the more complex it becomes it
makes a little bit more of expensive
solution and because of all those things
it's in to create a little bit more
overhead although you have fewer
administrators a little bit more
overhead managing that so that kind of
led to the next thing that we see a lot
of us what's called land free and this
is important the problem you'll see with
this with this to tape is you're doing
this transfer of data over your network
where you can create these conflicts
right so if it's you seeing network
you're doing normal network activity and
you're also throwing all your data
crossed for backup you kind of create
these problems that's why you see things
like land free so in this example using
something like an X and solution you see
that we map the drive on your ex or
frayed to an xserve all right so it's
now mapped is a regular dr your clients
hit that server access it
just like they normally would then when
it is a backup it goes through then that
network so you never touch this what's
called land free you never push the data
through the land you're always pushing
it directly from the server within your
fibre channel fabric great very very
fast in the most efficient right because
you're able to take advantage of this
really nice dedicated high bandwidth
communication get your data to the
appropriate area get the advantages of
central administration and it's a great
use of your backup resources right so
now you're using your fastest network
for your data and the most efficient
very some of the other ones a little bit
more complex a little bit more expensive
and more overhead your administration
but it's a trade-off right because you
get that better performance for your
data so as you see here we have a whole
slew of backup vendors and some of
obviously recognizable ones like a
legato of veritas that even great people
like backbones CA Quint x todas group
dance a tempo Archer we're avail all the
different vendors that we've worked with
so what I do here in this next slide
kind of give you an idea of the
different types of solutions that are
available so for example everybody's got
clients right and we have a lot of
vendors here now they see they see the
light they get it to get okay great
Apple has a lot of storage we need to
have more server solutions available on
the market so that's why you see a whole
slew of guys over here doing betas but
today if you need to divide basically
have three vendors to have a production
solution available today Vance Poulos
and SGL so I want to kind of give you an
overview here's the different types of
architectures here's why we got to be
able to back or dad and why it's
critical so now what I'd like to do is
introduce a few of the guest speakers
that I've brought when we look at the
market we kind of broke it down into
three basic groups we get at one end
where'd your cost is the lowest in your
complexity is the easiest its individual
workstation right this is your laptop
this is your desktop you know the need
to back up your own individual one and
then it comes a little bit more
expensive and more complexity you access
to small work group these are typically
departmental level or small group levels
it's like what we do in my own
organization I have my own little server
it backs up about five or six of us and
you know it's that workgroup server
solution so they need to keep a building
it's a really high end for corporate if
you're backing up terabytes of data and
you got it much information across
different areas all over the network
that's when you need it which leads me
to my first person like to bring up from
dance retrospec and I'd like to
introduce patli hi Pat welcome thanks
thanks very much appreciate it thank you
so dance was founded in nineteen
eighty-four we started out in the
Macintosh and we ship our first
Macintosh backup product in 1985 was
actually for an iomega product called
the bernoulli box with anybody remember
back that far we've been in the back of
the store industry for 20 years we have
to tus patented technologies one of
which I will talk about later and more
in depth and we've been working closely
with Apple for many years we've
initiated a number of server backup
programs of them are a number of apple
first over 3 million computers worldwide
have purchased licenses of retrospect
protecting their data today and we
partner with many leading hardware
vendors for example i mentioned i omega
earlier back in 1985 we're now partnered
with them today they announced the rebbe
drive their new low end tape replacement
one of our prices on a little bit and we
also bundled for example with the new
exabyte vxa two firewire library which
if you haven't seen yet you wanted an
easy connect solution go take a look at
it firewire interconnect perfect
solution for your xserve today and we
have a product line that ranges from
small home offices of a couple users to
large departments and you know a couple
hundred users the one thing we focused
on is making sure busy an easy to use
product most people aren't back of
administrators that are dealing with us
they want something they can install and
get up and running within a few minutes
most customers can install retrospect
start their first back up and start
protecting more data than five minutes
of install we think that's very
important in this market we're not
stopping being easy to use it's also
flexible we support a wide variety of
backup mediums we support tape which has
been the most popular backup media to
date for small medium business and we
support it whether it's connected via
firewire a scuzzy fibre channel you name
it we also support hard to
back up which is very key for either low
end or even high-end customers and for
customers who need off-site backup needs
we also offer ftp backup we have backup
signs of support Macintoshes whether a
certain system 7 max running power pcs
still all the way up through the latest
Panther clients Linux clients or Windows
clients one of the things that we focus
on is that intelligence we understand
storage you shouldn't have to understand
all the complications we should handle
this for you one of the things you have
is called progressive incremental backup
and what that does is it allows you to
say I just want you to back up this set
of computers and i'm going to say dubas
progressive back up the first time
retrospect backs up it automatically
backs up everything the next time only
backs up the newer changed file since
the last backup to a particular set of
media and the reason a particular set of
media is important is that you want to
have more than one sort of backup media
if you're relying on once the backup
media to protect your business and it's
in the building and something happens
you don't have a backup if it gives away
with your building going away you don't
have a choice so you need to have
off-site and on-site backup sets by
actually keeping track of your on-site
and off-site backups differently we can
give you complete protection for both
sets that's up-to-date and current so
you can restore from either set without
having to worry about it the other thing
that's intelligent about the approach of
progressive incremental backup is that
we have single instance storage
everybody in this room I look around I
see a lot of bright apple logo facing in
the eye from your power books and I
books that means there's a lot of people
out there whose data is not on a server
they need to protect it and the
important thing about that is there are
a lot of common files and use what
happens if we all have microsoft office
installed do I want to backup 10 copies
of microsoft office across the network
no I want to backup one copy and the
other nine I should keep track of so I
can restore them retrospect does that we
only backup one copy of every unique
file in a particular backup set it saves
you space and time and it makes it more
effective for your backup media security
again multiple baseline backups are key
if you don't have multiple baseline
backups that means if you're doing
incrementals against one full backup and
you lose that full backup your
incremental are useless against doing a
complete restore
multiple baselines are key and
protecting your entire network and
barefoot complete verification by
default we do bite for bite verification
of your data because if there's
something wrong between writing the data
to disk tape optical and we go back and
don't need a bite for bite there's no
guaranteed was written higher percent
correctly personally I was backing up to
disk recently and I had a bad firewire
cable on my personal backups the
verification pass caught the problem
with a bad cable and showed me the files
that were corrupt verification is key to
your backup strategies and tell the more
important thing about intelligence is
one of our patented features is called
backup server and the way to think about
this is its policy based backup you tell
retrospect what computers are most
important how often do they require
backup do I need to back up these
machines once a day once a week once
every three days you determine that what
day did you want to back up if you're
backing up there's no books you don't
want to back up their systems because
you can easily resist and reimage them
from your systems but you may want to
back up their users directories but
always exclude their cash and mp3 files
because those are not files that are
important to your business may be
important to them but it's not important
to running your business and you too
much backup media can be used and when
your backups can run whether they can
run all day or they can run at night
once you tell retrospect those things
retrospect will manage the rest we
automatically prioritize the computers
most of need a backup and batch them up
first so I'm on the on the road for a
week and I come back in the office and I
haven't been backed up I will
automatically be backed up as soon as i
connect to the network within a few
minutes it's going to find my computer
back it up because I'm the backup and
computer most in need if I'm rotating
between multiple sets of backup media
it's not a problem let's expect will
back up to the available media on the
computer at the time of the backup you
don't have to think about do I have the
Monday tapes here or the Wednesday tape
here you just put in the media you want
to back up to and let respect will
manage that in another key important
fact about this is if a problem occurs
during a backup a LAN segment goes down
somebody turns off their computer or
anything like that retrospec well note
that backup didn't complete and
prioritize that computer for backup at
the top of the queue so next time it
finds the computer on the network it
will start backing up again
and it will continue the back up from
where it left off so you have to recopy
everything it's really key and managing
your backup time this automatic error
recovery and moving forward on your data
well we're talking about backups today
it doesn't matter about backup Chris's
story told you a lot and you'd be able
to restore the data and retrospect was
designed to restore your data the most
common restore we see from years of
doing this is that people need to
restore a few files because that's the
common case oops I made a mistake i
overwrote my budget excel file I've
committed to save how do I get this data
back it's not a problem that respect
allows you to search for a wide variety
of criteria I can find all excel files
created by Chris in the last two weeks
and it will go and just find those and
all my available back at media easily in
one place it's very key to help you
narrow down your search for your
important files but when disaster
strikes that you care the most about
restore nobody ever wants to think about
it but it happens retrospect has what we
create snapshots which are basically an
image of what the computers like every
time it's backed up so every day
assuming with retrospective progressive
incremental backups the first day do a
full backup and then we do incremental
backups every day since then we still
keep track of all the files that have
been moved renamed delete it on the
computer at every backup stage so if I
need to restore my computer after
multiple incremental backups to three
days ago because I either deleted some
system I got a virus or I had an issue
like that you tell retrospect restore to
that point in time three days ago you
know will restore that computer to
exactly that point in time at the time
of the backups so if I moved renamed or
deleted files they're not going to come
back and I restore if you've been doing
restore some other traditional products
which do full of incremental is that
don't track the state of your computer
when you do the restore to that point in
time you're going to get that file as
you moved renamed or deleted you don't
want that you want you to computer to be
back to the exact point in time it was
and to help you aid in that worst case
scenario retrospect includes a bootable
zest recovery CD that you can boot off
the computer up a CD do the restore and
make your life easier without having to
first reinstall the system so thank you
very much for your time tomorrow we'll
be in the enterprise lab I tlab you have
any questions from nine to two pi
remember correctly it's like breakfast
nope anymore and thank you very much
thanks that so now what I'd like to do
is introduce our next featured speaker
who from our toeless group that they
target their solutions called brew
answer it back up for story Philadelphia
does agree makes sense kind of name for
your product and you can see that
they're targeted pretty much in the
middle although almost all of my vendors
here can scale from their simple to the
high end I was trying to get them you
know to look at okay what's there what
they're known for and so without further
ado Tim Jones hi kam welcome thank you
hey are we there did I break it okay um
brew is yeah we've got a good back up
some brew has been around for going on
close to 19 years now we have been a
strong contender and a strong player in
the UNIX market space because that's
where we came from we grew out of
Motorola technology in the middle early
80s and we've worked their way onto just
about every UNIX platform that you can
imagine and if any of you guys have got
a UNIX platform that I don't currently
have a copy forth you'll give me an ssh
login in 30 minutes we're there we we
picked up OSX and started supporting it
in 2001 based on some requirements of
our existing customer base we have
solaris users and AIX users and hp-ux
users and Cray users that we're bringing
OSS OS x boxes into their environments
and they said hey why can't we use brew
to back those up too so we got the calls
and we started taking care of it our
team does one thing we do backup and
restore we don't do crystal are we don't
do database reporting we don't do games
we don't do word processors we partner
with every major tape and library vendor
running out there some of them you know
something you don't know that there are
just literally millions of existing
licensors how many people have ever
use an SGI box well if you've done a
system backup you've used brew how many
people have ever used Motorola's back in
the old days when they were using the
88,000 well you were using brew some of
the older Linux installs from folks like
red hats and those guys included brew in
the actual operating system and again
like I said all we do is back up one of
the most important elements that Chris
hit on is reducing that back up window
how many of you have four days to
perform one backup we got one hand back
there I didn't think I was going to see
many hands there simply because we don't
we need to get the data into its backup
environment as rapidly as possible and
be that stage on disk is Chris mentioned
or onto tape whether it be a standalone
or a library environment we need to
reduce the amount of time that that
operation requires the thing that brew
brings to the table here is we provide
very high throughput rates and very low
CPU overhead primarily a good example we
have users that are constantly using
brew and multi-drive scenarios where
they're backing up to and streaming data
to multiple drive simultaneously and on
a 1.33 gigahertz xserve our numbers look
like load factors of about point four
with cpu utilization of about twenty two
percent what this means is that while
that backup is going on you're actually
able to continue doing useful work we do
our verification quite differently for
most others every 2 K of data that we
pull off of your file system we apply a
32-bit checksum to the result is when it
comes time to do a verification of the
data that's on the tape we only need
brew and the tape we no longer need your
file system so for those of you that are
doing live back up with live
verification we just sliced your backup
window in half we incorporated just two
disc capabilities directly into our
current Bruce server product some of you
have come by I see some faces out there
that I recognize when I cover the light
and one of the things that we've done is
we've added this technology right into
the product native you don't have to buy
it as an add-on because we realize again
we've got to reduce that back up window
to help you be successful in your
administration tasks by writing to disk
two things happen first of all as Chris
said you're getting just to the speed
your only limiting factor is your
network infrastructure but in addition
if you've got lots of clients out there
if you're backing those clients up to
take those clients are having to take
turns you're backing up one client it's
finishing your backing up the next
client is finishes and so on with a dis
stage environment all of your clients
can backup asynchronously limited only
by your network bandwidth if you're
dealing with systems where you've got
100 or 200 megs of data that changes on
a daily basis an average incremental
backup can take as little as 10 or 15
seconds so when you when dealing with
staging so when you look at it from that
perspective all of a sudden now this
work group stage backup window has just
reduced itself dramatically finally as
Chris mentioned the next thing to do
with stage data is to put it onto media
that can then be stored either safely
on-site or off-site as Pat was
mentioning the key there is again how do
I do this without interfering with my
daily work schedule well because bruise
upstaging occurs between simply the
server on which the server application
is running and your tape library again
the rest of your network backup
environment is not involved at all so
you're not you can do this at your will
it doesn't matter and again because of
bruges low fingerprint and our footprint
and low overhead even the server that
we're running on does not recognize a
serious impact with that operation
finally on the upstage when we upstage
that data from your stage disk to the
tape drive we actually replicate the
data into a tape backup as if you wrote
it originally to the tape this means
when it's time to restore our clients
data it's a one-step restore you're
taking the data from the tape and
returning it to the client there's no
need to restore the upstage environment
to the stage disk and then restore to
the client we are about ninety two
percent POSIX our core engine is just a
whopping 290k on OSX the eight percent
that's there that's not POSIX applies to
supporting OSX tape I o OS x disc record
utility which will be coming in an
upcoming release the utilization of
special features for some of the special
os's such as PTX and I Rick's and these
things where you have to take into
account the way you allocate memory for
certain things these are the little
pieces that are specific to the various
OSS but that makes up a little eight
percent of what we do this means this is
why I was not joking when I said if you
can give me a login in 30 minutes if I'm
not on your platform we will be we
provide unique mechanisms to prove the
backup as I said every two king of data
that comes off of that file system is
checksum when it leaves the file system
because between the time that at least
the file system and the time that it
hits the input buffers on your tape
drive there are a number of places where
corruption can occur because we check
some at that early we can tell you very
rapidly whether or not you've got a good
backup and we can tell you down to 2k
within an existing file on tape whether
that data is any good the key here is in
the event we run into a problem take
media flakes how many of you have never
had a tape problem take media flakes
alpha particle alien abduction coffee
spill I you know nobody's figured that
one out I've been doing this since 1984
and if somebody can tell me I'll be
happy to help fix things but because
we're able to do things the way we do it
in the event that we run into a problem
on tape we continue to seek well first
of all we try to reread that problem
area five times before we continue and
we seek past that point to the next
logical header on tape and we continued
restoring your data so even if you run
into a situation if any of you have ever
used tarted will back them tried to do a
store and been greeted with something
that said tar take bio read error well
you're not going to see that with brew
well unless the tape really is
absolutely bad and it can't be accessed
finally cross-platform i mentioned all
the platforms that we're on the one
other thing that's important about that
is because we utilize the same tape io
layer for all of the platforms you can
take tapes that you backed up on SGA SGI
you can restore them on OSX or you can
send them to somebody with ASCO box or
you can send them over to a linux
platform or a solaris platform or any
other platform and we will allow you to
restore that data intact we take account
things like big Indian versus
little-endian byte swapping if you've
ever used cpio because you needed to
move data from a big Indian system to a
little Indian system you know you have
to add special Flags brew recognizes
those changes and takes care of all that
for you so when you get into that area
we've taken care of it we are highly
extensible hey that should have been
highly Christmas we are highly
extensible via scripts and compile
compile wrappers brew as i said is 290 k
you're able to take that and combine it
with whatever language you want to use
cocoa carbon java TCL TK python with TK
inner real basic it's your choice you
can put your own wrapper around brew and
make it do things that you want it to do
but you don't have to worry about how we
get the data to the tape or back from
the tape or to the stage environment as
developers if you're interested we can
talk to you because we actually provide
that 290k kernel as an OEM product and
we can help you build back up into your
application brew understands and
supports all UNIX and mac OSX elements
now how many of you saw the tiger
keynote how many of you know what an ACL
is well guess what brew 17 dot 0 which
is the core of our current application
platform does not support Tigers ACL
however in about two weeks it will we
are thank you we are compatible with OSX
or less than 15 through the latest stuff
now I did a naughty and I install tiger
on my laptop and we'll see but I have
very good feeling that we are not going
to be facing any problems there we left
the UNIX layer of OSX to provide that
small footprint and high performance
with low overhead we utilize the scuzzy
standards the standards for tape SSC and
for changers smc to ensure that whether
you're using the smallest firewire
device or the largest fibre channel or
scuzzy library we've got you covered we
don't also charge extra for those types
of pieces so if you've got for libraries
you're going to pay us one price for a
product and like I said we support all
scuzzy all fibre channel all firewire
and USB devices natively no additional
drivers so and i also forgot of course
one question that comes up yes we cover
your resource Forks and all of your
finder info so if it was positioned here
on your desktop when you backed it up
when you restore it it's positioned here
on your desktop so that's it thanks very
much and unfortunately I'm not going to
be available tomorrow at the lab but you
guys can visit us and Chris we've got
information on the web for how to get in
touch with us here to be into the show
all right great thanks for Thank You
temperature alright so now you could see
some great solutions from Hanson toeless
group so now we're going to do is look a
little bit more at what we call the
high-end corporate and market and to do
that we have backbone even though I
spell does name wrong sorry about that
its v8 k but anyway I'd like to
introduce today Andrea bulls the senior
director for strategic alliances or
backbone thank you Chris welcome let you
down
thank you this is a partnership that
began almost a year ago today I was
approached by Apple and they sense a
growing demand in their core customer
base many of you have been with Apple
for some time an explosion of data is
led to requirements that just simply
weren't there before requirements that
range into the multiple multi terabytes
and certainly we've seen that over the
past few days seeing that and the
visitors have come by and the
conversations I've had in the hallways
more and more customers are looking for
us to help them reduce the complexity so
want to put a timeline here and part of
apples moving towards a UNIX base is
certainly in line with where we come
from our heritage is out of Bell
Laboratories so we've got strong UNIX
heritage when Chris mentioned companies
that are crazy enough to support 50 60
different operating systems that's us
we're all about customer choice and one
of the challenges in working with with
Apple was understanding a market without
delving into a huge business case it's
just not part of apples cope corporate
culture so we I went in I developed a
case along with Chris and the any Apple
team and I'm proud to say that success
to us and we announced our beta look
look to me like 30 customers maybe 30 to
40 customers would be part of the
backbone that vault beta today we have a
hundred and twenty customers signed up
we've got major research laboratories in
Europe we've got two of the most
prominent universities in Texas one of
the Ivy League universities actually
wrote us a p.o based on screenshots and
what that tells me is a there's pain out
there they want they want an alternative
to there's excitement there's a lot of
excitement about what Apple's bringing
to the table in terms of xserve raid in
terms of X serve it is absolutely killer
product I've got engineers here on sites
that are seeing some of the exterior for
the first time enter enter pumped up so
that's that's a lot of fun for us you
know we're proud to be a part of it
Steve Jobs featured backbone in his
keynote and I after I have to say I grew
up and I followed this band around the
Grateful Dead and
and I don't know of a more loyal
following you know you look at something
that back in the day that Jerry Garcia
would espouse and certainly people would
follow it and when Steve Jobs put
backbone up there in the January time
frame during is keynote 25 calls the
next day 25 in how do we get part of
this where is the beta a lot of
excitement on our on our end beta is
active so part of my call to action to
each of you today is please visit the
backbone site bak hbu any free beta
program its ongoing and also includes
support for oracle 10g so that's a big
focus for us is not just a standard
backup but we have an application layer
and are available to support 10g awesome
mysql so big focus there for us you know
again when we built in that vault we
built it back in the early 90s like a
lot of products however we took a hit
and in the mid 90s 96 97 we rewrote it
from the ground up to take advantage of
networked storage environments a it's
got a very easy to install GUI driven
product but also it has a modular
architecture an architecture that allows
us to design and develop very quickly so
in the case of oracle 10g 40 s pen we're
talking about weeks of development not
months like some of our standard
competitors veritas and legato and the
incumbents have a much longer rolling
development cycle disc the disc you've
heard about it a lot i'd be surprised if
most of you aren't experimenting the
next step is taking a look at a virtual
tape library virtual disk library and it
allows your disk device to look like a
tape target when Chris talks a little
bit about the growing compliance needs
this is a way to begin setting yourself
up for that a way to migrate off the
tape and do it effectively and also
naturally as you heard recover quickly
so again I think you know if I was
asking you to take a few things away
from this robust this two-disc
capability and also a broad choice in
the application side you know again we
don't see that there should be a
trade-off moving to OS 10 we want
give you the ability to run oracle run
mysql and do it on your time i want to
put a sample graphic up here just what
it what a standard we call that acne
engineering might look for again hot
backup that would be your Oracle MySQL
we've been doing this for a long time
and you know I will also want to point
out we've supported freebsd in the past
the first was a pretty clean port you
know I point to a customer that I spent
a lot of time with down the peninsula
and that's yahoo they chose their a
large freebsd shop they run a lot of
mysql for those of you that aren't sure
whether mysql is ready for the
enterprise environment i'd urge you to
think again i think many of you be
surprised it's coming a long way coming
fast that's 18 data centers across
multiple continents and again their run
it on freebsd so i think that apple and
moving forward with OS 10 has picked a
winning winning platform and you know is
staying with their mantra which is all
about reducing complexity from a
backbone and apple standpoint we share
that as a common vision taking more and
more complex environments and making
them symp you know easier and easier to
implement one thing i want to come you
know for you to come away with is a
weird company that's easy to do business
with we have a sales model that's
entirely indirect so you know whether
you want to pick up a partner in the
form of apple and purchase met ball
through the istore or if you have a
common VAR that you're already used to
doing business with so i'm going to take
care of your needs we're there to enable
them back one of the company is entirely
indirect we work with our channel
partners we work with OEM which is apple
and key is vs which is Oracle MySQL
sybase and others so again i'm looking
forward to the Q&A portion of this
session encourage you to throw some
questions our way and again want to
thank apple and chris for having us here
thanks a lot
[Applause]
good job thank you already that number
our designated hitter I guess you could
call actually it we've also invited
another one of our vendors a tempo who
also has a solution targeted up here
with their product called time navigator
so I'd like to welcome Randy Batterson
director of strategic alliances come on
Chris all right thank you well I'd like
to kind of kick everything up and just
give everybody an idea about who a
temple is and what we do basically the
company get this right here I know it's
going to do that first time for
everything but basically a tempo has
been around for 12 years the company was
founded in nineteen ninety-two we're
actually duly located we've got a
facility in Paris France we also have a
facility in palo alto california right
here in the bay area we actually
consider our product a best-of-breed for
high-performance enterprise environments
the product has been engineered from the
ground up it's very very high-performing
if you look at our customer base it's
actually a global 1000 customer base we
have clients like France Telecom huge
telecom company been with them for
almost seven years and they've got about
13 different sites around the country
itself and what's really interesting is
that we actually support almost 3,000
clients from a central location on that
particular facility so again it kind of
goes back to that enterprise you know
functionality it's a very very robust
product we've got almost 2,000 customers
worldwide and out of those customers
we've got roughly 3,000 sites and I
would say a good chunk of those are all
very high-end performance or high
enterprise type companies and if you
really look at it some of these
customers have a very high computing
requirement and that's again reason that
they go to this product because we have
a strong suite of business continuity
and compliance solutions to offer him
the company itself has about 170
employees right now we're growing
I rapidly we've got about 40 for
technology people focused just on our D
so this is really kind of shows the
commitment you know that we have towards
developing new products just like
supporting mac OS x and you know we've
put a lot of emphasis on development and
i think this kind of shows that most
companies don't have that type of a
ratio of engineers focus just on
development some of the industries that
we support include telecom finance
manufacturing education government a lot
of the key industries that are also
touched upon with the mac environment so
again this is a nice marriage between
you know what we're doing supporting mac
OS x and the industries that we clearly
support so we think that we're going to
have to really get additional e okay I'm
just going to dive in a little bit here
about what the product is a time
navigator one of the things that we like
to say is it really it's got the fastest
restored period and that's if I can
leave you with one thing today just
remember that restore is what we do best
and we do it really well there's some
really key features that help support
this kind of going back to the name time
navigator we do something which is
called time navigation and the way we do
this it's actually the ability to go
back in a point in time any given point
in time we can actually go and retrieve
a file or directory at in any instant in
the past so it's a lot like going back
into the past you can actually pick on a
particular date time and bring that file
back and we do this in a unique way we
use a an object oriented database and
why this is unique is if you look at the
way many other solutions actually
perform this function they actually do
it with a flat file database and what
that causes is when you go and actually
do a metadata search you're actually
going in a sequential form so you
actually have to go and you know go down
the line and actually find the
information looking for just like if you
were searching on a tape by using an
object oriented database or relational
type database it's instantaneous so that
search takes a matter of seconds
nanoseconds I'm going to fly in at boom
you're there what's really nice about
time navigation to is the way we
actually display that information and
what we do is we put it in a kind of a
tree structure so it's very
user-friendly it looks just like what
you're looking at in any type of a
directory tree so if you're looking for
a particular file you put the date down
that you're looking for this file and
what happens is when you click on it it
does that search instantaneously it'll
bring back that tree and it will show
you exactly the way your data looked at
that point in time and what's really
nice is it will actually show you with
little crosshairs on a particular file
if that file is deleted or there was a
change to that at a point in time now
where we get a little more deep in this
whole certain time if you will is we
give you the ability to do a depth of
search so you can actually click on that
that hatched file click on it and it'll
show you every change that it's ever
happened to that file and then what you
can do at that point is actually click
on any one of those changes and bring
back that image or the data the way it
looked on that particular time so it's
very quick it's very unique in that
sense you're getting a you know a point
in time visual if you will or the way
your data looked and then you can
restore it so it's very quick synthetic
full backups this is another thing that
we do it's kind of like what I guess
what we like to say that it's an
incremental forever what if you can
imagine when you go in to do your first
backup you're going to have to do a full
backup okay after a point in time what
you're going to start doing every day
every day if you set up your policies
what you're going to do is necromancy
laughter that and then another X mental
another if you were to do say a one-week
set up and at the end of the week so you
started your backup on Sunday you
finished on Saturday what's going to
happen is offline time navigator will
take those incrementals and it will
build a new full it's a full out of all
those incremental this can go on forever
and that's why we kind of call it up a
forever incremental paradigm it keeps
doing this forever you never have to do
a full again and you can add points you
know delete what happened before so you
don't keep accumulating all that that
baggage where this is really unique for
the restore why it makes it fast is
going back to that database what you're
able to do is if you let's go back to
the scenario where you did your your
full
and let's say after that first week you
want to go back excuse me and you want
to actually find a file on Wednesday
what happens is time navigator because
of the database and being a relational
type database will go exactly back to
that particular file on Wednesday so
what you don't have to do which happens
with many other applications is you
actually have to go and load the full
then go back through you know the sunday
full then the incremental Mondays and
their middle Tuesday just to get back to
that Wednesday file so it makes it very
fast you're not having to go through
that whole cycle of going through tapes
another thing that we do is Mackel
multiplexing macro multiplexing is
really nice in the sense that if you're
familiar with multiplexing the way it
happens today basically you're going to
get you know multiple tape streams
coming back to the backup server and
what happens on the backup cycle is the
data is actually laid down onto the tape
and little little bits and pieces when
you go to do a restore you're actually
having to go back and pick out little
points of that that file information and
fit it through what happens is the tape
drive actually starts doing a shoe
shining effect and this actually slows
down your restore what time navigator
does is we actually create a buffer and
when the backup job kicks off the data
is put into 256 megabytes pockets okay
and these buckets then what's the the
backup starts you know feeding into this
buffer it will go up to like a watermark
and when that water marks hit the backup
job will kick off and the data will
actually start get laid down onto the
tape and what happens it gets laid down
at least 256 megabytes well when you get
to the restore instead of having to go
back and look for each and you know each
little piece of data like you would with
regular multiplexing / interleaving you
actually back up using these big chunks
and they come right back to the server
very fast basically hitting the theater
okay theoretical throughput of the tape
drive not getting a lot of support there
another thing that we do is actually
this is quite unique we've got a pointer
here you guys see that kind of play all
this point one of the cool things that
we do actually on a security standpoint
is we're able to back up through a
firewall using just one poor many
applications today actually require
opening up multiple ports as you can see
over here on the left-hand side what
happens Thanks actually what happens is
yes and these other applications you
have to open up multiple ports many for
to initiate the back up and then many
for you know bringing the data through
on the other side so you have anywhere
from ten to twenty five ports opened up
what happens is a lot of these ports
remain idle and that opens you up to the
threat of any kind of security attack
with time navigator we actually only
open up one port that's an outbound port
opened up by the administrator so
everything from the metadata to the
actual data flow to the TCP actually
goes through that particular port and
again that's an outbound port so it's
not a it's not open from the outside
it's open from the inside inside the
firewall so nobody can actually
penetrate from the outside this is
really cool too because it there's a
couple very large deals that we got just
to give you an idea capgemini ernst
young you know auditors I mean these
guys really worried about protecting
their data right one of the things that
they picked us over or pick us over the
competition if you will it was because
we had the security feature and it was
very key for them to pick us because of
that ok time navigator just a few more
innovations that we have end-user
simplicity you know really cool thing
that we do is a data center services
model the data center services model
actually is something that we have that
allows a user to kind of rent it
rent-to-own if you will it's kind of
like leasing a car instead of actually
having to go out and purchase the
product in its entirety all up front put
up a big
capital expenditure we're allowing
customers actually kind of pay-as-you-go
will go in we do an assessment of the
customer site I understand what their
data requirements are set up kind of a
billing cycle if you will for them based
on what we think their usage will be and
then they're actually able to you know
match against that to every quarter and
see kind of where they're at and pay on
a monthly basis at the end of a period
of time like a year or two years if they
want it can actually purchase the entire
thing or you know just keep using it on
a monthly basis so it's a real nice way
to actually get into backup and restore
especially an enterprise environment
sort of laid out a lot of cash they can
do it very cheaply multiple backup
streams up to four this is really a very
unique feature that we have also so if
you can imagine if you're backing up to
say an x rayed and you want to kind of
invoke some dr features and stuff that a
lot of people are doing in oh we've been
talking about dt dated cape this
actually gives you the ability to back
up to four different devices at the same
time so the same data stream could be
backed up a seriously to four different
devices and it could be any mix of
devices it could be taped disc mo and it
could be any number so you could have
two disc one tape one mo and where this
comes into play is for example if you're
going to have some kind of a dr scenario
so you're backing up to the x-rayed and
you want to backup to disk locally just
have sup them into near line at the same
time you want to have something offline
and that would be say an off-site dr
site you can do those asynchronously so
if one breaks the other one's going to
go regardless and we could do it up to
four so if you want to have four
different sites we can do it just to
just tape I touched on that one really
unique thing that we do is just a
district 8 we actually are well we
enabled a customer to do this actually
automatically if you will we set
watermarks and so you're able to migrate
the data automatically from disk to disk
the tape and it's really a nice feature
because you know if you reach a certain
point watermark will kicked off after a
certain period of time you can set it by
duration or whatever and you'll have a
fully automated
mr. Tate solution so people are looking
for you know ilm type solutions this is
something that would really help out and
flexible API CLI basically API set we
touch on every feature in the product
this is really key to I mean if you've
got different you know products within
your environment and you're not sure if
their support or maybe we don't support
them through our api's it's very easy to
write and support the product set CLI
you know command-line interface same
thing if there's some scripting that
needs to be done to support a product we
can do it very easily so that's about it
but one thing if I can leave with you
today is that time navigator basically
is the fastest restore period and that's
something that will you know challenge
anybody on we've got some really good
features and we think they're great so
thank you very much thank you Rani so
you see some really excellent solution
from a number of vendors that support
the mac OS 10 platform so just kind of
tie it all up again to kind of repeat
what we talked about so we looked at
initially some of the storage trends
it's driving the market today everything
from compliance to the need to reduce
your backup window and that's driving
your strategy for how you need to back
up your systems we looked at a number of
different backup architectures obviously
just regular backup accessing your man
doing disk disk to disk to tape to land
free and then as you see we have a
wealth of back of choices available
meaning any me you would have in your
particular environment and last but not
least I'd like to invite you tomorrow
morning in the enterprise IT lab I'll
have everybody here we have some demos
showing some of their equipment any
question you have on backup I'm sure
we'd be able to answer it there tomorrow
at the enterprise IT lab so if you have
an opportunity for me please to stop by
and see us and we'll be glad to have you
all right contact information for myself
Pat Tim Andrew and Randy and I'd like to
ask my four colleagues to come on up and
join us here