Brian Peterson is a storage architect for a Midwestern Fortune 100
consumer packaged goods manufacturer (you know the name, but
Peterson said he would rather we not mention his company in this
article). The massive environment at the company's headquarters,
where Peterson works, consists of 500 terabytes (TB) of EMC Corp.
primary disk, 300 TB of VTL, 1200 McData Corp. and Cisco Systems
Inc. ports, Symantec Corp. NetBackup and IBM Tivoli Storage Manager
(TSM) running backups, Symantec Veritas Volume Replication and
EMC's SRDF for replication … whew, is that everything?
What's your most important storage project for 2007?
Brian Peterson: We are constantly working on improving DR
[disaster recovery] recoverability. Even if the data is replicated
off site, testing should be continuous. We are also planning to
define and implement a records management policy. We need to
prepare now for the lawsuits that may occur five years from
now.
What do you think will be the biggest hurdles to implementing
storage in the coming year?Peterson: The next hurdle we will have is how to manage
the information we store. We can store the bits and bytes on disk,
but with hundreds of terabytes, how will we manage
information in terms of retention and expiration? Increasing
regulation without viable technologies and methodologies to help
will make life more difficult. We've figured out how to scale, but
really we don't know what we have.
What storage technologies are you evaluating for 2007? What
looks interesting to you?
Peterson: Records management tools, improved backup
storage (something better than tape or VTL) and storage capacity
planning tools.
Have you seen any kind of backup alternative that looks
promising? Why doesn't tape or VTL seem like the best approach?Peterson: We like tape and VTL because of the compression
on the drives. We get up to 4 to1 compression, saving lots of
storage resources. The VTL improves upon some of the traditional
limitations with physical tape libraries. With VTLs, we can create
hundreds of virtual tape drives. Each server can have its own
dedicated drives, reducing the complexity of shared tape
resource.
The SCSI tape protocol has some downsides shared by both VTL and
physical tape. Both are very sensitive to latency in the [storage
area network] SAN network. Neither enables concurrent reads from
the same volume by two separate hosts. This means tape replication
for DR and restores cannot happen at the same time. Finally, tape
drives are just finicky when it comes to SCSI resets and
reserves.
I haven't really seen any good alternatives, yet. I'm looking
for disk-like resiliency and flexibility with compression or data
deduplication. It should allow multiple hosts to have concurrent
read access to the same backup image. It should be scalable beyond
a single frame without inducing additional management overhead. A
guy can wish, can't he?
What do you predict will happen in the storage market in
general in 2007? What technologies will be the most
important?
Peterson: It's hard to say what will be the most
important development this year. Data deduplication holds a lot of
promise and could radically change the way we store and retain
information.
What do you hope WON'T happen in 2007?Peterson: I hope that we won't run out of disk and have
no money to buy more. Or let a 100 TB disk array fail. Yikes! I
hope we don't get sued and have to produce all the spreadsheets
touched by executive X in the last three years.