Server Firmware Upgrade Highlights
- Most leading vendors are restricting firmware upgrades to servers under warranty
- Some doubt over whether critical fixes will be supplied to those without
- Designed to protect supplier investments and IP
- Limits customer maintenance and disposal choices
- Changing server platforms is harder than buying a new car
- Could herald a return to Vertical Integration and customer lock-in
Back in 2010 the EC forced auto manufacturers to share their technical data with independent garages. Many thousands of car drivers in Europe have experienced a wider choice and cheaper servicing since then. Unfortunately car manufacturers are not the only ones to want to exert more control over their products and customers – you’ll want to know more about something similar happening in the IT industry.
Restricting Patches To Servers Under Maintenance
The data centre market has been hard going for a number of years (see Figure for the sluggish revenue growth of new servers, storage and networks), which has increased supplier competitive and defensive activities.
New computer servers are typically covered by a 1-year hardware warranty; for the rest of their life-span (typically 6 years) service contracts are provided either by the original vendor, a third party, or occasionally by the users’ own technical staff. However a number of suppliers are now preventing users without maintenance contracts from receiving firmware upgrades. If only the initial purchaser with an up-to-date maintenance contract will have the access to future patches, these changes will limit the use of third party maintenance, second-hand equipment values and trade.
Which Vendors Restrict?
We believe the following policies are in place at the leading server vendors:
- IBM – ‘Fix Central Machine Code updates are available only for IBM machines that are under warranty or an IBM hardware maintenance service agreement’, ‘Starting in January 2014, IBM will implement entitlement validation on Fix Central for select software products and updates and for Machine Code (also known as firmware or microcode) updates for select machines…’; these rules apply currently only to its Power servers
- HP – announced at the end of February 2014 that it ‘we will provide firmware updates through the HP Support Center only to customers with a valid warranty, Care Pack Service or support agreement‘, although in a later clarification it said it wouldn’t make the changes retroactive and would continue to supply safety and security patches for free to all (see below)
- Oracle – ‘You will be required to have a valid Oracle Support contract or Hardware Warranty before being entitled to obtain access to firmware updates’
To date we believe there are no similar restrictions on servers at Lenovo, Dell or Cisco (which interestingly does appear to have a similar restriction for its Routers and Switches). Could IBM’s statement that ‘these entitlement validations are not being implemented in all countries at this time’ (ibid) indicate nervousness about the legal status of this move in some regions? I’m certainly not sure, but all vendors and governments need to be aware of the implications of such moves.
Why Are Vendors Doing This?
Mainly to make you buy new servers. Many users have migrated from Unix to Linux on x86 (see Figure) due to the advances in Intel and AMD chip performance and the lower cost – and ease – of maintenance, making things worse for the ‘alternative’ Unix/RISC/Itanium server vendors, who need to continue to invest in their IP despite this loss of revenue. One way to do this is through high margin hardware maintenance services. However this frustrates many users as the price usually rises and utility (including the number of patches) decreases as the equipment ages beyond 3 years. This why many turn to third party maintenance – something limited by the new contracts.
In the ‘alternative’ server area each vendor (IBM, Oracle and HP) has a migration programme in place for moving customers to their own platform.
As part of these activities they have revised their reseller and distribution relationships to support these solutions. They will argue that users will be more secure if they use authorised resellers and partners for buying and maintaining their servers.
There has also been a strong move in the last few years towards Converged Infrastructure and Integrated Systems (CIandIS), where servers, storage, networking and infrastructure software are combined in solutions designed for speedier workload deployment and greater business agility. While these depend on ‘secret sauce’ software for the orchestration of southbound components, they also require more sophisticated (and more frequently updated) firmware than standard ‘horizontal’ rack servers. Hardly surprisingly then, these kinds of system too are also subject to the ‘firmware upgrade only under maintenance’ restrictions. Last year some users we spoke to thought the vendors had rushed their products out before being fully tested, which I’m sure resulted in frequent patching.
So you can assume that the more expensive, proprietary and integrated your server is, the more restrictive the contract will be. However vendors may apply these restrictions to x86 servers of all sizes in future. HP, for instance, recent announcement covered the entire x86-based Proliant family, although it subsequently clarified that it ‘will continue to provide … [security and safety] updates to all server users’. IBM’s restrictions (see above) don’t cover System x – but Lenovo might conceivably introduce them once the acquisition’s complete. Oracle’s restrictions already cover x86 alongside Sparc servers.
Can’t Users Just Pick Other Servers?
Whether changes to ownership contracts are big issues or not depends on how easy it is for users to change from one type of system and vendor to another. The Single Unix Specification Version 3 (SUSv3) standard in z/OS, AIX, HP-UX and Solaris ensures software portability, but interestingly is not supported by Linux operating system vendors. In practice very few users would succeed in swapping between alternative (or from alternative to x86) platforms without having to spend large sums in other areas like storage, networking, middleware, testing and even hiring new staff to manage the system. In addition software licenses, server, storage and network components need to be renewed at different times, which limits the scope for change.
As always things are easier for new as opposed to upgrading workloads – one small service provider we spoke to last year installed an integrated system in 2 days, but then spent 9 months swapping over each customer infrastructure. It’s also telling that server vendors’ competitive wins often involve extensive ‘proof of concept’ demonstrations and lengthy implementation engagements.
Over all changing server platforms is much more challenging than picking a new car. By introducing their new contracts vendors are restricting the number of third parties that could help and increasing the cost. Users have to think in terms of ‘time to implementation’ alongside ‘return on investment’.
Some Conclusions – Vendors Choose Between Matrix or Vertical Integration
Increased integration shows itself in other ways of course – the growth of Cloud Computing, Apple’s success, the shifting emphasis from Cap Ex to Op Ex are all aspects. Data centre vendors are preventing each other from servicing equipment, severely limiting the market for Multi-Vendor Support – another aspect of advancing integration.
It’s all being going on for some time.
I proposed a 3-stage multi-decade view of supplier strategies back in 2010 in which the Horizontal phase – created by the anti-trust case against IBM on-going from 1969 to 1982 – gives way to the Matrix one. In my view Matrix differs from Vertical in its dependence on Open Source components and the vendor’s keenness to avoid locking its customer in.
The ‘patching only under warranty’ rules are only a small part of the changes in vendor approaches, but they suggest we may just be returning full circle to Vertical Integration.
Users need to be aware of potential lock-in, unrestricting vendors should make more noise and government bodies should at least consider these changes. If the EC made it easier for car drivers to use independent garages to protect their citizens, they should certainly have a look at this aspect of the IT industry as well.
I’m very interested to hear from all suppliers and users about this important issue. So please contact us at info@itcandor.com or comment below.
Having been a leader in the OVMS, AIX, Microsoft, Red Hat, Solaris, SUSE and Ubuntu operating system management and AlphaServer, Power, HP 3000, HP 9000, Itanium, SPAR, X86 and X64 server field for over twenty years… all the contracts with manufacturers have not changed and always excluded updates and patches for customers not under service contracts… so the recent announcements are really a restatement of facts. It is interesting to note that HP is unique in their offering of operating system Software Update Services (i.e. patches and updates) either bundled in with hardware and software service contracts or ala carte for those customers who feel comfortable in their expertise and/or availing themselves of 3rd Party offerings. Perhaps we should encourage all manufacturers and Independent Software Vendors to follow HP’s example and offer hardware patch and update services for hardware as they currently do for operating systems. Our strategic alliance partners have worked as friendly partners to both manufacturers and 3rd Parties to satisfy each customer’s specific requirements.
Excellent work. I completely concur that once availability and pricing of repair is monopolized, the OEM not only gets all the high-margin service contract revenue, they also dictate when they will drop service and thereby guarantee new sales.
It is poorly understood that this destroys the value of the equipment because no one will purchase used equipment if they cannot get it repaired. Lack of a used market destroys the collateral value of the asset for banks and lessors. The machine eventually has so little value that it shouldn’t even be considered an ASSET for accounting purposes.
This is far worse than most appreciate, and not at all the norm since the dawn of the computer era. Hardware patches and fixes used to be delivered as Engineering Changes and more recently as downloadable firmware patches and fixes. (now called “Updates”) Fixes to design errors have always been part of the OEM’s obligation to deliver products that work up to specifications. Mixing availability of corrections to design errors with other functions does not make them any different. OEMS should be separating their defect support from paid enhancements and not playing word games. Hardware patches should never be combined with OS patches.
I hope the discussion continues.