Victoria - Fotolia


Survive the server orchestration takeover

The tenets of traditional IT -- well-tended data center hardware, owned software deployments -- are falling away in the cloud era of standardization and orchestration.

Building a hybrid cloud, the architecture favored by almost all enterprise players, requires a different mindset to IT organization.

With a focus on standardization, the cloud hardware department is a much simpler place to manage compared with IT shops full of mainframe, mini and even traditional commercial off the shelf (COTS) servers. Hardware and infrastructure admins have something to worry about as server orchestration and other intelligent automation replaces many of their previously differentiating skills.

Cloud presents the issue of scale. Most successful cloud computing infrastructures live on large blobs of identical servers. These servers evolve rapidly compared with traditional IT infrastructure, so the blobs are not homologues of each other. Even so, the server groups are fundamentally identical in being x64 architecture, with all the peripheral component interconnect express, serial advanced technology attachment and Ethernet interconnects structured roughly the same way over generations.

Homogeneity in the cloud extends from the server to network hardware and storage appliances.

The new IT order

Admins will exercise tighter control through governance policies embedded in the server orchestration, a way of automating management tasks. At the same time, they can allow business units the flexibility to spend their IT budgets as needed in the cloud resource pool.

Software-defined storage takes this one step further and simplifies all of the storage platforms, migrating the burden of administration off IT pros and onto the software and systems side. Once storage is set up in a big pool of shared resources, data center orchestration performs most routine jobs without an admin's assistance.

To add to traditional job woes around orchestration and automation, consider that many bigger shops will adopt a no-repair policy whereby cloud hardware is overprovisioned in lieu of warranty. In a no-repair deployment, a disk drive failure is fixed by transferring to a spare drive, not removing and replacing or repairing the failed drive.

Is there anything to relieve the gloom for the hardware team? Their platform-level skills will be in demand during the period of churn that server orchestration and systems automation involves. They will need to figure out the new structures as well as support the old. The survivors will become more sophisticated and have in-demand skills.

Your future IT job

In the long term, server and overall data center orchestration will move into the platform territory very strongly, with consequential downsizing.

Admin productivity is often measured in petabytes per admin or some similar key performance indicator. These goals are okay for static environments, and worked well for the past three decades in enterprise storage, for example. The notion breaks down when structural shifts of the magnitude of cloud server orchestration occur.

The 10-fold increase in hardware capacity in just four years illustrates why set per-admin measures are weak. It probably takes just as much effort to control an older 4 TB array as today's 40 TB compressed storage array. It's true the software feature set for the new product is more sophisticated, with more efficient interfaces, but not 10 times more so.

So, a hardware and platform admin faces a bleak future, but software roles may expand to fill the void. If you are an expert in building mash-ups, you've little to fear. Similarly, enterprises will seek big data, in-memory database and analytics expertise. Cloud administration and system tuning will become more important as IT shops realize that successful hybrid clouds require broad knowledge in how to identify and fix bottlenecks and inefficiencies.

Software administration is under a different threat -- the addition of rent (software as a service, also called SaaS) as an option beside make and buy. On-premises enterprise resource planning software is succumbing to SaaS in the cloud options. Even Microsoft Exchange is looking to the cloud for a future, with the majority of mailboxes expected to be in the cloud -- Office 365, Google Apps -- in 2016.

The hybrid cloud will require skilled software pros for all these new programs, and at least in the short term software admins will see a surge in demand. The long-term outlook isn't so good, since the SaaS vendors will pull in much of the work. By spreading the same software development and management effort over a large client base, they take advantage of a high performance/task per admin metric.

So far, orchestration and software as a service present a bleak picture for the traditional data center admin, but there is a possible untapped source of demand for admin skills. Over 400 billion lines of COBOL are still running worldwide, often in mission-critical applications. Reality is setting in, and the move to COTS, x64 and the cloud is beginning in the legacy space. The transition means more work for legacy admins, though the end result is cutting them out.

About the author:
Jim O'Reilly is a consultant focused on storage and cloud computing.

Next Steps

Most of these changes are relatively slow, with five- and 10-year timescales and a lot of resistance to change. For any admin thinking about the future, picking the right career path and being in the right place are important. Some admins need to evolve their skills, while others might want to branch off into new areas.

Dig Deeper on Scripting, Scheduling and IT Orchestration