 |
Dr. Geoff Barrall
CTO, BlueArc Corporation
|
|
Dr. Barrall is the CTO, executive vice president and co-founder of BlueArc Corporation and the principal architect of its core technology, the SiliconServer Architecture. Prior to joining BlueArc, Dr. Barrall founded four other ventures, including one of the first Fast Ethernet companies and a successful UK consultancy business. In this role, he was involved in the
introduction of innovative networking products into UK markets including the Packeteer and NetScout. Dr. Barrall received his PhD in Cybernetics from the
University of Reading in 1993.
|
|
|
Much has been said about the process of storage consolidation, but for a large geographically distributed enterprise it has been easier to discuss
than achieve. The benefits are obvious in that if a corporation can bring
most of their storage back to a central location, the cost of managing that
storage goes down and the utilization of that storage goes up.
However, there have been two key issues with achieving this goal.
The first challenge on the path to storage consolidation was that of
creating a storage pool large enough to meet the demands of a geographically
distributed enterprise. Today, the increase in disk storage density and new
fast fiber technologies have allowed storage vendors to create storage
products (EMC's Symmetrix, Hitachi's Thunderbolt and BlueArc's Silicon
Server) that are at last able to support the speed and scalability required
to meet this need.
The second hurdle has remained with every enterprise since the early days of
WAN deployment. However much empty fiber does or does not lie buried in the
ground, the ability to move large quantities of data between geographically
disparate sites still remains elusive, caused in part by the cost of large
data pipes and in part by the unavoidable latency in any long distance data
transfer. Anyone who has tried to access data stored at other geographic
locations in their enterprise has probably experienced this kind of
frustration for themselves. The problem is especially disruptive for
enterprises like Broadcom who have a number of regional engineering offices
with a requirement to share and interoperate on design data between sites.
Trying to do this kind of distributed design over a wide area is certainly
frustrating, if not impossible, today.
These kinds of bandwidth and latency issues are analogous to the early days
of the Web. In order to speed up Web transfers, companies like Inktomi and
Network Appliance created caching appliances that could hold copies of Web
pages and were placed (geographically speaking) much closer to the end user. This allowed for faster page retrieval and an overall better browsing
experience. Today, we take this kind of caching as a given in our Web
browsing experience.
The challenge of using technologies like these Web caches for enterprise file
caching has been around for a long time. However, enterprise file caching in this manner involves a far larger problem than
Web caching: Enterprise file protocols like CIFS and NFS are much more
complex than HTTP, and data needs to not only be read (as Web pages are) but also often needs to be written as well.
Despite these challenges, a new generation of products is emerging from
companies like Tacit Networks and DiskSites that, for the first time, allow
file caching to be as transparent as Web caching. These new file caches will
fetch files from a centrally-located storage pool out to remote offices
where they are installed allowing the users there to read, modify and create
files with the same kind of performance they would see if they were at the
same location as the storage pool. These clever caching devices also offer
compression technologies that make it much quicker to move modified files
out of the cache and back to the main storage pool.
Through the installation of a central cache manager (located with the main
storage pool) and sophisticated distributed data management schemes, the
file caches allow for users at remote sites to modify files without any risk
that users on other remote sites or at the head office will be modifying the
same file at the same time. Logically, the file never leaves the main
storage pool.
Some of the benefits of this type of technology are obvious, in terms of the
enhanced speed and usability they bring to distributed file operations.
Other more subtle benefits may well change the way distributed office data
storage is designed in the future. By effectively allowing fast file access
to head office data, file caches effectively remove the need for smaller NAS
or file servers to be distributed at remote offices. This reduces the capital
expenditure and management costs required for these devices.
Another bugbear
for system administrators managing remote office data is that of backup. If
no qualified IT users are present on the remote site, then how can that
remote data be ensured to be kept safe? Again, the file cache removes this
issue by storing files back to head office consolidated storage pools where
they can be backed up as part of the normal data protection routines there.
All in all, it seems that the new breed of file cache appliances are likely
to change the way that smaller remote office data storage is handled in the
future, as well as enabling a new wave of distributed data sharing and
cooperation by allowing more and more data to be stored in a centrally
consolidated data pool.
This, hand in hand with the new generation of Web
services, should make the headache of distributed data management a lot
easier to deal with.
Copyright 2002, Blue Arc Corporation.
