I’ve been a bit of a lurker watching the development of the NHoS project, and have spoken to a couple of you about it a number of months ago. I’ve got some observations to share, and a potential solution to the problems that arose towards the end.
Firstly, a bit of background - I’ve been involved in deployment and management projects in the recent past where my team has successfully deployed thousands of Linux systems, remotely, over marginal bandwidth, and on demand. More on that later. Personally, I have something like 35 years of experience in IT, with the last 18 spent almost exclusively working with Linux based systems - Intel, Power and ARM based.
We are all aware of the failings of Windows as an OS, and also pending removal of support of Windows 7. But at the risk of (unintended) insult, I feel that yet another flavour of Linux is almost certainly not the answer. Allow me to explain why.
The biggest problem of software is legacy: maintaining it after the original creators have left the organisation, lost interest or worse.
Introducing a “NHS flavour” of Linux is a laudable goal, but in my opinion, focuses on the wrong things. Massive deployments of Linux have ultimately failed in the past for just those reasons: some enthusiasts did it themselves, the people involved moved on, the project lost momentum, and was ultimately binned. (Munich and Lower Saxony have both followed that trajectory, for example.)
All the operating system is, at the end of the day, is a launch platform that provides authentication to get you to the applications. Nothing more: nothing less. The color, name and font are irrelevant if it doesn’t run the apps.
The questions that need to be asked are:
- does the operating system support the software we need it to? (*1)
- does the operating system support the hardware we already have? (*2)
- can the operating system be professionally supported by it’s vendor(s)?
- does the operating system have the appropriate security certifications for risk management? (*3)
- is the code open source and unencumbered(*4)?
- are there multiple vendors of the software?
If the answers to most of those questions is yes, then you have the beginnings of a solution.
It is likely not to be possible in any organisation to adopt a 100% open source model, especially one as complex as the NHS. But we can go a long way by using standard, supported Linux operating systems such as Ubuntu, Suse, and Redhat. Add to those standard OS bases drivers such as the excellent work done on smartcards, portals to legacy systems via RDP and Citrix, and modern application access via HTML5. (*5)
Oh, and you can make it look even better with an NHoS-like theme as well.
The key factor here is the final support is provided by a third party - reducing risk.
Having multiple vendors reduces risk even further.
As an added bonus you can choose the correct OS for the job rather than the only one.
Back to Windows 7: the maths of an upgrade project are astounding. By the time you add the cost of acquisition, downtime, installation, and disposal of the old system, it will cost up to £1000.00 per system. For 1000 systems, that’s £1 million - just to maintain Business-as-Usual (BAU) operations. No added benefit, unless you count Windows 10 as a benefit…
Realising this, we’ve developed a system that perform zero-touch, in-place Linux deployments on existing hardware. Planned correctly, it takes minutes, not hours - and at the end, a user can log directly into the AD/LDAP directory immediately, using their existing credentials, to use the applications that they need according to their user group. (*6)
In that 1,000 machine example, you’ll have saved around £925,000 over the cost of a Windows 10 upgrade. Then you work out how to use those savings to convert some of those legacy Windows apps into something that will run anywhere - not just on Windows.
Though not specifically developed for health care, we realise how it might transform NHS systems. We are very interested in trialling in an NHS setting to test our assumptions. If anyone is interested, drop me a note here, or for more information, head over to https://requiredmagic.com and tell me what you think.
Thanks for reading!
(1) - doesn’t have to be native support. RDP, Citrix, HTML5, VNC and even X can be bent to do what you need it to do.
(2) - hardware life can be extended beyond the vendors own warranty period, but engineer the infrastructure allow for it’s failure and/or reuse.
(3) - it might be cool to have a custom-compiled kernel, but that increases risk because the vendors won’t support it.
(4) - it’s a nice goal. Start with the client OS and work outward to the servers.
(5) - Even the HTML5 version of Office365 works on Linux.
(6) - For example, who really needs Office? Maybe 20% of users. Group them together and eliminate the costs of everyone else.