This article can also be found in the Premium Editorial Download "Storage magazine: How to successfully deploy data replication."
Download it now to read this article plus other related content.
There's no question that people with different skill sets within IT view the problems and their solutions differently. "Network engineers have a different point of view for I/O [than storage admins]," said Tom Becchetti, senior storage and Unix engineer at St. Jude Medical Inc. in St. Paul, Minn. "The only time the network folks are concerned about latency is when there is a distance involved. Most of the time for local traffic they are only concerned about bandwidth. Merging networking and storage doesn't work."
For example, "before I arrived here, they [the network and storage groups] decided to use a dedicated NetApp filer with an iSCSI attachment for [Microsoft] Exchange," Becchetti said. "The network connection, with a separate VLAN [virtual LAN], was part of the entire network infrastructure. The disk response time was so bad, it caused outages for Exchange. In many shops, a network I/O isn't monitored as closely as a disk I/O. Long story short, they converted the attachment to Fibre Channel and the problem was solved. The network group stated that the network was fine."
Becchetti's position is very storage-centric. According to Stephen Foskett, director of the data practice group at Mountain View, Calif.-based Contoural Inc., "network and storage pros have critical insight into their areas," and continuing to isolate them in specialized silos "would be a tragic loss akin to what
Foskett added that new technology is changing the status quo. "Right after the data center is virtualized and IT infrastructure is recombined, applications themselves will fundamentally transform, demanding a merger of the current IT infrastructure and IT applications groups," he said. "This could all come within five years, or it could be delayed or diverted by organizational infighting and intransigence."
DBA jobs are changing, too
The roles of database administrators (DBAs) and their working relationship with storage administrators are also changing. Mike Shapiro, distinguished engineer at Sun Microsystems Inc., agrees: "I'd like to see the conversation between DBAs and a typical storage group get beyond 'I need a LUN of size x gigabytes,' which is how it works today."
|How many terabytes should a storage administrator manage?|
Just how much storage should a single administrator be responsible for? It depends, especially in an interdependent IT environment.
James Damoulakis, chief technology officer at GlassHouse Technologies Inc., Framingham, Mass., said this isn't a particularly meaningful metric. Of course, as disk capacity grows, the number continues to change. More importantly, the biggest factors relating to TB/admin are the complexity and dynamism of the environment. Very complex environments -- those with lots of different kinds of devices, complex architectures and so forth -- by their nature require more people to manage them. Likewise, a fast-changing environment with many provisioning requests and high volumes of data movement also increases staffing needs.
"These are the areas we always look at first in terms of making staffing efficiency recommendations," Damoulakis said.
This was first published in April 2009