Migrate your Alpha environment to another platform
I read an interesting statement on the business2community.com website a week ago; it said “The pyramids in Egypt and the Great Wall of China were built to last forever. Computer hardware and software? Not so much.” This seems like a good place to start when talking about (in this case) upgrading your AlphaServer systems.
You may remember I published a blog a few weeks ago entitled “Why should Alpha customers be thinking of migrating their systems?” The driving forces behind this blog were the issues associated with owning AlphaServer systems today, the big ones being:
- The manufacturer (HPE) no longer supports the hardware
- You cannot access modern storage arrays with the available storage adapters
When you get situations like this that are “out of the norm”, this generally translates into paying extra money for support, for sourcing devices that work with the old hardware, and so on. On the other side of the argument, you often get “if it ain’t broke then don’t fix it”, so let’s us look at the pros and cons of upgrading your AlphaServers.
- The system is working fine - and has been for a number of years.
- You have 3rd party support on the hardware – “we have a 3rd party company that looks after our systems; they are doing a great job; why change?”
- You have maintained the version of OpenVMS at 7.3-2 because it is stable – “we have been using OpenVMS 7.3-2 from HPE for years and it has never let us down.”
- You do not have the in-house resources to carry out a major upgrade – “to be honest, our OpenVMS resources have been dwindling recently as the staff retire, and we do not have a program for replacements.”
- The CFO is not keen on capital expenditure on new equipment at this time – “because everything is working fine, the CFO does not see the rationale for buying replacement equipment. We have an annual shot at getting the expenditure into the budget and if we haven’t seen any problems with the system the CFO defers the expenditure until the following year.”
These are all perfectly valid reasons for hanging on to old systems. I am not sure I’ve heard every one of these reasons from a single customer, but I have certainly heard all of them at one time or another.
Let me put a different slant on each reason:
- The system is working fine – I cannot argue about that, but one day it will fail and the question you need to ask yourself is how much that failure is going to cost in terms of lost revenue and customer confidence.
- You have 3rd party support on the hardware – again, a perfectly valid argument; however as time goes on genuine spare parts for old systems become scarcer and more expensive, and the engineers to maintain the systems retire.
- You have maintained the version of OpenVMS at 7.3-2 because it is stable – true, OpenVMS 7.3-2 is stable, but do not confuse stability with support. OpenVMS 7.3-2 cannot be supported at L3 (engineering) by HPE or anyone else, so if there is a major issue with the operating system, it cannot be fixed; this includes any latent security issues in the operating system and its layered products, or any new security patches which require implementing.
- You do not have the in-house resources to carry out a major upgrade – I am hearing this rationale more and more recently and it is definitely an issue for some companies; however this obstacle can be overcome. There are companies who can provide OpenVMS resources on a project or contract basis; obviously VSI can do this (other brands are available), so this is probably not the barrier it might seem.
- The CFO is not keen on capital expenditure on new equipment at this time – this is a tough one. I have had many a conversation with a CFO where capital expenditure (or lack of it) has been the crux of the conversation rather than the need to acquire new hardware or software. Nowadays it is much easier to purchase these items on an “as you go” basis. Most software can be purchased on a subscription (even OpenVMS) and all hardware vendors provide leasing options. This takes away the CapEx versus OpEx argument and lets the CFO plan a smooth budget without any capital expenditure hiccups.
Do not get me wrong, I thought AlphaServers were the bee’s knees, but they went EOL in 2007, so unfortunately their time has come and gone. While I hopefully refuted the cons, there are actual pros for upgrading your AlphaServers:
- Get access to the latest technology - replacing your AlphaServers is a great opportunity to invest in the next generation of devices with greater capabilities. An example would be moving to an X86 platform; immediately you have the opportunity to utilise cloud technology, or maybe you want to get involved with the Internet of Things.
- Reduce your costs - running AlphaServers in a DC (or computer room for those of us of a certain age) is relatively expensive in today’s currency. When they first came out AlphaServers were considered very efficient, but 14 years after their demise probably not so much.
- Speed up processing and get more done - slow systems may be a symptom of a deeper, underlying issue in your IT infrastructure. Upgrading the infrastructure will address these issues to help you deliver faster results, and faster applications mean a more efficient business.
- Get on to manufacturer-supported hardware - AlphaServers went EOL in 2007 to be replaced by Integrity servers, yet there are still hundreds maybe thousands of AlphaServers still in operation today. This is testament to their durability and their ability to deliver; however HPE no longer supports these servers or most of the devices that connect to them.
- Get onto the Cloud - in a recent report (June 2020), Gartner predicted an increase in public cloud usage of 50% between 2019 and 2022; this translates into an increase in public cloud spending of $121 billion. This does not mean that the cloud is for everybody or everything; however it does demonstrate that a very large proportion of companies world-wide think it is a useful technology for their business.
So, there are a number of good reasons why companies should be looking to get off AlphaServers; however those reasons have existed for many years, so why are there still hundreds/thousands of Alphas still in service? It could be complacency; it could be the Alphas are doing such a great job (why fix it if it is not broken); it could be costs; who knows?! It reminds me of the situation when you hesitate to take out insurance for that treasured item; it is the same kind of argument – the watch is working fine why do I need to insure it – oh, we had a break in, and they stole my Rolex. We have all done it (not the Rolex bit; that was just an example), where everything is working fine, and we see the insurance as an unnecessary expense. It is exactly the same with older computer systems; they are working, why do I need to upgrade? So it runs your production line; how much does it cost you if your system goes down and the production line stops? Unfortunately, the older the equipment gets, the more prone it is to go down.
Let me finish with a cautionary tale. Back in 2018, computer security experts discovered a virus called Spectre. Spectre is a particularly nasty type of virus; it works by tricking a program into accessing innocent seeming memory, but actually allowing an attacker to read this data and potentially retrieve sensitive information without user approval. It appeared that every type of computer could be infected by this virus, so VSI carried out its own research and the conclusions appeared in this customer letter VSI Spectre customer letter.
Fortunately, through good engineering, OpenVMS on Integrity and pre-EV6 Alpha systems was found to be immune to this virus, and we think it is the same for EV6 and later Alpha systems. Looking forward, the X86 port will mitigate the effects of Spectre on OpenVMS on X86 systems. Hang on, I just said “we think”; that is not good in technology terms! The reason I said ”we think”, as explained in Eddie Orcutt’s 2018 customer letter, is that VSI have not been able to reproduce the behaviour in those systems. This does not mean the problem will not occur on those systems, just that we could not reproduce it in a lab.
Let me be clear; I am not saying that all EV6 and later AlphaServer systems are prone to this virus, but I am saying that there is a chance, under the right circumstances, that your system may be susceptible (to this and to other similar such security threats that may arise in the future). Operating in an unsupported environment leaves you with nowhere to turn for help.
Any questions, you know where to get me.