Storage jobs are changing, and storage professionals who know storage and nothing much else are going to have a thin time of it in the years ahead. Any idea that you can stick around in a comfortable niche, knowing about RAID levels, LUNs and SAN fabric topologies and nothing much else are going to get blown out of the water by convergence and IT stack integration.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
Having said that, storage array controllers are getting more complicated and capable, but their management and integration with networks and hypervisors is becoming deeper. Enterprise SAN arrays now come with thin provisioning, tier data automatically across solid-state and hard disk drives and carry out replication and snapshots, meaning their internal environment has become vastly more complicated than it used to be.
Data storage jobs affected by convergence
These systems' ability to do these things automatically once policies are set and to carry them out at the direction of upper-stack-level software such as ESX through APIs means the storage admin needs at least a working knowledge of the storage status, reporting and management functions of ESX or other hypervisors. They'll also need knowledge of the networking environment that connects networked storage to servers because the departments running applications will want answers to storage data access issues and couldn't care less if these are due to hypervisor/storage links, fabric or disk array problems. They just want their data to be available.
Canny storage professionals will recognise that the storage silo is having its walls removed. They will also recognise that new skills are coming into play. One of the most obvious is Fibre Channel over Ethernet (FCoE). Storage staff will need to understand how Fibre Channel signals are carried over Ethernet and how Ethernet itself has been modified to provide the lossless transmission needed by Fibre Channel.
Meanwhile, networking staff will need to understand things like TRILL, which provides the routability that multi-hop FCoE networks need. These matters are complicated by it being early days, with different vendors bringing different implementations of FCoE and TRILL to the market. But FCoE is coming, as are TRILL-type functions, and these had better be on your skills hit list if you're involved in storage.
Data storage jobs and server virtualisation
Virtual servers are established as a fact of life, and it's vital to understand VMware's storage- and backup-facing functions, such as vMotion and the vStorage APIs. Storage professionals will also need to understand the VMware side of the storage interactions in arrays that are integrated with ESX, and I reckon those who understand both sides of the VMware/storage coin will become very valuable. This applies to both block access and file access storage arrays.
With the rise of unified storage, it will become harder for block or file specialists to remain in walled file or block gardens; any employer that buys a unified storage system will need admin staff that can work both sides of the file and block divide. Object storage is still pretty much a discipline on its own, and, with Dell pushing its DX6000 object storage platform, there will be more demand for people who can make it run through the hoops needed by its users.
Both VMware-related skills and unified storage knowledge are examples of how elements of the overall IT stack are converging. The big story here is the advent of completely integrated IT stacks as single orderable entities, such as Vblocks from the VMware-Cisco-EMC coalition and equivalent systems from HP. This type of uber-product could grow in popularity, and it will need to be managed as an entity and not by separate people looking after separate functional parts of it. This will also spur the need for new skills on the part of storage professionals.
Data storage jobs and leading-edge technologies
A whole new area of expertise that is coming along is solid state disk (SSD) and its use as server cache, storage array controller cache and storage array Tier 0 data storage, along with automated data movement such as EMC's FAST or IBM's EasyTier. The likelihood is that SSD use will spread and its management become integrated with the management of devices using it. Where externally attached flash SSD is used -- or DRAM from Texas Memory Systems and similar suppliers -- there is a niche storage management speciality area that should endure as long as the suppliers do.
Another niche area, but a rather bigger one, is in the detailed management of enterprise storage arrays such as IBM's DS8700 Turbo, EMC's VMAX and HDS's USP-V. These have gained more and more functionality, with automated data movement being the latest. Setting these arrays up and keeping them optimally tuned for their end users is a continually demanding and enduring job.
Federated storage is another area that's starting to look interesting within and across data centres. This is leading edge, as are cloud deployments of systems like EMC's Atmos, which may or may not become a major EMC product line. Atmos skills may or may not pay off. 3PAR skills, however, surely will pay off now that HP is set to buy that company.
Now let's turn 180 degrees for a moment and look at skill sets that will be of negative worth and should be wiped from your CV and relegated to the history section. These mostly relate to vanished or rapidly vanishing media formats. Raise your hands if you can manage Sony AIT or SuperAIT tape systems. You can? Learn a replacement skill set at once. Ditto VXA, Mammoth and other disappearing tape formats. The safe formats for employment purposes are LTO, DAT, and the IBM and Oracle/StorageTek proprietary ones.
Lastly, anyone qualified to look after Plasmon optical storage should be prepared to enter the museum of lost job opportunities. Retrain at once!
About the author:
Chris Mellor is storage editor of The Register.