Select the directory option from the above "Directory" header!

The mainframe computer is not dead: CA expert

The mainframe computer is not dead: CA expert

"Cloud is the catalyst for people to get a grip on what’s core to their company and what’s not"

Reports of the death of the mainframe are exaggerated. While all the attention given to Cloud computing - generally delivered from large-scale server farms rather than ‘big iron’ - may appear to suggest that the lid is on the mainframe’s coffin even if it is yet to be nailed shut, there is a contrary view based on economics.

Distinguished engineer and chief architect at CA Technologies mainframe business unit, Scott Fagen, says one of his goals is to help the company’s existing mainframe customers to become “more agile, more Cloud-like”. This contrasts with CA’s more general role of helping customers decide what they should use.

He makes the point that there is a crossover point where doing IT internally is cheaper than using an external provider thanks to declining costs per MIPS (millions of instructions per second) due to economies of scale with large systems. Furthermore, the latency involved when two systems talk to each other within the same data centre is lower than when communication is over a distance, so it makes technical sense to co-locate certain functions.

Fagen also suggests that IT should be considered and managed in much the same way as a supply chain. There’s nothing very controversial about his idea that core functions are best kept in house, while others may be usefully outsourced. But what he does make explicit is that the context of a function is as important to the decision as the function itself. He points out that where some retailers are happy to rely on external logistics providers, Wal-Mart runs its own fleet of trucks because it sees logistics as a core function.

One of the most commonly quoted examples of non-core IT is email. The standard position is that this is a generic service that can be safely outsourced, but Fagen suggests that it should be considered core if potential leaks could be highly damaging to the organisation.

There’s also the question of timeliness. While losing access to for a few days might be inconvenient, it shouldn’t be a major drama, especially if the sales pipeline is relatively long. But losing an HR system for more than a short period could cause serious problems.

Fagen’s point is that such considerations are “not going to hurt the mainframe side of the business” (ie, of an organisation’s IT operation) because we have already reached the stage where all mainframe workloads are classified as core, with all the others being migrated to other platforms, which may include the cloud.

“Cloud is the catalyst for people to get a grip on what’s core to their company and what’s not,” he says.

Fagen also notes that running Linux on System z is “cheap” as long as cost accounting is being done correctly. For example, one bank discovered that the entire data centre power and cooling costs were being applied to the mainframe budget. That was appropriate when the centre was commissioned, but over the years a wider range of equipment was installed yet the cost allocation didn’t change. After a thorough review, the bank concluded that the actual mainframe costs were an order of magnitude cheaper than it previously thought.

And he says the low cost of running Linux on a mainframe has led Wells Fargo (a US bank and financial services business) to engage in a number of ‘re-insourcing’ projects.

Agility is an important part of the cloud promise, and Fagen says CA is working on tools to help internal mainframe operations match the flexibility of cloud providers. One example is that the company’s AppLogic cloud computing platform - currently x86-only - is being ported to z/VM, and will allow the easy movement of Linux workloads between x86 and z/VM systems.

Mr Fagen suggests the questions should be ‘what’s the most viable way to deliver your services?’ and ‘how can you quickly redeploy services if a cheaper way emerges?’ CA, he says, is building an ecosystem to help IT organisations understand what they’ve already got, where they need to go, and where the crossover points between different technologies fall, plus management software to help internal IT operate in a more cloud-like manner.

Stephen Withers interviewed Scott Fagen at CA World 2011 in Las Vegas, which he attended as the guest of CA Technologies

Follow Us

Join the newsletter!


Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.

Tags cloud computingLinuxmainframeyARN

Show Comments