A mainframe-based cloud service?

As an industry analyst, it feels like I spend hours of my life sitting through the same old cloud, mobile, social and big-data stuff from IT vendors. Sure, the world is changing and technology is developing – you’d need to be living under a rock not to know that. But once a marketeer gets that clicker in their hand, there’s no stopping them telling you all about it again. So you sit through the obligatory 5 to 10 slides on industry trends waiting for them to get to the point and tell you what, exactly, they are offering to customers.

On the odd occasion, though, someone breaks the mould. This happened recently at an industry analyst roundtable discussing the future of the IBM mainframe following its 50 year anniversary.

It started when someone stepped up and declared they had built a cloud hosting business around the mainframe rather than going down the more obvious x86 infrastructure route. It’s not often that a room full of analysts all come to attention simultaneously, even though some of us looked around for the people in white coats coming to lead this crazy man away for some much-needed therapy. Yet when that man, Lubo Cheytanov, started to lay out his numbers and business plan, I was reminded of the thin line between madness and genius.

Cheytanov’s reasoning became clear as he told his story, which provided some interesting insights into how to think a little differently about TCO in the context of cloud services, and how you need to consider software licencing and maintenance costs, as well as the infrastructure side of things.

When Cheytanov conceived his cloud service operation, L3C, he naturally considered a traditional x86 based foundation for his venture. During the planning and budgeting process he calculated what he would need in terms servers, floor space, cooling capacity, electricity, software, manpower, etc on the basis that the business would ramp up reasonably quickly.

Being an ex-IBMer, though, he figured that if you are going to build for scale from the outset, you might as well run the numbers based on a platform that was originally conceived for high levels of capacity and throughput, i.e. the IBM System z (a.k.a. ‘the mainframe’). In order to make the comparison between the x86 and System z meaningful, however, he then looked at the most important metric for the customer, Total Cost of Ownership (TCO), and did the arithmetic in relation to commonly occurring workloads.

An opportunity that emerged from this analysis was the provision of Linux VMs (based on z/Linux) to support Oracle database and middleware consolidation. The reputation of the mainframe for high scalability and resilience was judged to be an advantage in this area, but more important were the potential cost savings. The logic was that based on Oracle’s current licensing model, you can get more throughput per license/maintenance dollar spent on the mainframe than on x86. Pulling all the numbers together, Cheytanov’s analysis told him that the ‘bang per buck’ differential was significant, and that this was relevant to lots of potential customers.

Of course not everyone would agree with this analysis, and debates have raged in recent years about the relative merits and costs of z/Linux vs x86 (e.g. running VMware) for Oracle workloads. However, Cheytanov was convinced of the arithmetic and this, together with the size of the Oracle installed based, was a key factor in basing his operation on a System z foundation.

Wind the clock forward and an Oracle consolidation service has now been launched. The offering is called COZy – Complete Oracle on Z Service – which is aimed at enterprises and ISVs that have a requirement to host around 100 or more VMs running Oracle. At this level, the expectation set (on the L3C website) is of a 20% saving on Oracle maintenance fees when migrating existing applications, and “instant ROI” when deploying new licences. You’ll have to do your own calculations to verify the claims as there are a lot of variables to consider, but it could worth exploring if you have a big dependency on Oracle.

While Cheytanov says that the COZy concept formed the core of his initial business plan, if you are going invest in a mainframe, you might as well exploit it in as many ways as possible. L3C has therefore developed a number of other service offerings.

One of these is ZEUS, which is primarily aimed at providing resources on demand for the mainframe community; a high end play primarily aimed at supporting z/Linux testing and development activity. Addressing the lower part of the market, L3C is then offering E-Z, an e-commerce platform running on the LAMP stack. Finally there’s Z-RO, which serves up Linux VMs on a wholesale basis to VARs, SIs and anyone else who wants to go down the MSP route but can’t justify (or don’t have the appetite for) building their own infrastructure.

Time will tell whether Cheytanov ultimately ends up in the madness or genius category. In the meantime the L3C story highlights how a bit of lateral thinking can potentially open up new opportunities for hosting companies and their partners.

Real business is more likely to get done if you stop waving your arms and singing along to the industry tune, and offer something more focused and tangible.


Registration required

Click here for more posts from this author

Dale is a co-founder of Freeform Dynamics, and today runs the company. As part of this, he oversees the organisation’s industry coverage and research agenda, which tracks technology trends and developments, along with IT-related buying behaviour among mainstream enterprises, SMBs and public sector organisations.