Opinion on the VMware debacle: Finally rely on open standards!
Companies are looking for alternatives to vSphere and ESXi because they don't want to pay Broadcom's prices. It's their own fault, says Martin Loschwitz.
(Image: Shutterstock/Igor Golovniov)
- Martin Gerhard Loschwitz
The champagne corks must have popped recently in Bräuhausgasse in Vienna's 5th district. VMware had just announced that various previously free components of vSphere and vCenter, including VMware ESXi, would have to be paid for separately by customers in future. As a result, VMware bills are exploding in many places, without there being any technical added value worth mentioning in return. Accordingly, horror stories are piling up in the relevant online forums from companies that are suddenly having to pay 500 percent or more of the original amount for their virtualization.
This is more than just a reason for Proxmox, which has its headquarters in Vienna, to be happy: Proxmox is something of a logical alternative for those companies that need virtualization but don't have enough workload to justify building their own IaaS platform. The people in Vienna probably sensed that Broadcom's decision had considerably improved the prospects for their own business – and have since been busy expanding their own partner network as well as Proxmox VE, the actual virtualization product.
What is often overlooked in the anger about VMware and Broadcom is the fact that a migration from VMware to Proxmox is also a real upgrade in another respect: While VMware is a thoroughly proprietary product, Proxmox is based on open source components such as KVM, Qemu and Ceph. This also means that should Proxmox ever try to repeat VMware's stunt, it would be easy for affected administrators to replace the product with alternatives. If push came to shove, basically any Linux distribution on which Qemu, KVM and Libvirt are available would suffice. This applies to almost all established Linux distributions. Operation and administration would usually be less convenient than with Proxmox VE – but if the continued existence of your own company is at stake, administrators will be happy to accept this.
Told you so
My first professional stop in IT was the Viennese company Linbit in 2006. It is behind DRBD, a kind of RAID 1 over the local network. DRBD has been part of the Linux kernel for a long time. It enables the construction of highly available storage systems. Combined with other open source components such as Samba or one of the various iSCSI targets, DRBD is a cheap replacement for SAN or NAS appliances on standard off-the-shelf hardware with hard disks that do not have to be paid for with unicorn powder.
This was precisely the core of Linbit's sales strategy at the time: that customers would not be dependent on a single provider for either hardware or software if they built their own central storage based on DRBD. Open standards and open source were and are the guarantee that manufacturers running amok will not drag their customers down with them when they celebrate their own demise with relish. People who have emerged from self-inflicted immaturity – one would be inclined to believe – would be aware of this fact even without further explanation; after all, it is self-evident.
Fast-forward to 2024: NetApp is as popular as ever, companies are facing bankruptcy because VMware is making a central component of their IT infrastructure unaffordable. And like lemmings, companies and even government institutions are making a pilgrimage to hyperscalers, which have taken the principle of lock-in to a whole new level. Because once you have migrated your own infrastructure to AWS, Azure or GCP, you don't just move it back from the cloud to your own hardware or even migrate it to the hyperscaler competition. Such projects require huge amounts of effort and money, which many are unwilling – or unable to afford. It is clear where this will lead: hyperscalers will gradually increase their prices and their customers will pay for lack of alternatives.
It makes you want to scream with rage when institutions such as the Federal Employment Agency proudly announce that they have successfully made the switch to Microsoft Teams. Or when German Chancellor Olaf Scholz personally intervenes in favor of the Delos cloud, whose backend is Microsoft's Azure. The fact that the head of government of the world's third largest economy is almost begging to be allowed to transfer even more money to Redmond via detours is an outrageous farce. Especially for those who have to pay for the fun in the end, namely German taxpayers.
Videos by heise
A rethink is needed
It's not as if no one had warned those in politics and business. For decades, representatives of the open source community have been repeating that only free software and open standards can provide a stable basis for infrastructure and, in particular, critical infrastructure. For some time now, Kurt Garloff has been promoting the Sovereign Cloud Stack, which enables customers to choose freely between platforms from different providers and, if desired, to migrate – or to operate a corresponding platform themselves.
There are proprietary products and various free alternatives for practically every basic component in today's data center, which are often better and offer more functions. Of course, from an IT operations point of view, it takes more effort to find and implement the right alternative from several free alternatives through testing. More effort, in any case, than having a sales drone gently sprinkle you with PowerPoint slides for hours on end until you even believe that you had the idea to purchase the proprietary solution yourself. In 2024, however, the convenience of those responsible can no longer be a valid argument for technical bungling. Anyone who handles their IT job in this way will have to vacate their position. "Nobody ever got fired for buying VMware" may be true, but it is – once again – part of the problem.
Instead, there needs to be a widespread realization that open standards and free software are not an option, but the only way forward for IT infrastructure in the long term. Especially when it comes to public infrastructure that serves the general public. Companies and institutions must be willing to get their hands dirty in the beginning in the interests of digital sustainability, instead of just trusting the colorful brochures of the providers. The fact that their salespeople in many places are downright pandering to IT managers should no longer obscure the fact that they are primarily concerned with their own bonus for signing a contract – and not with the sustainable success of a company. And anyone who is still digesting the VMware disaster because they themselves are affected by it should urgently go in search of other Damocles hanging over their own set-up and tackle their elimination as soon as possible.
VMware customers are in luck: providers such as Proxmox are pulling the coals out of the fire for them this time, not without publishing the resulting tools under a free license. It is by no means certain that the next time it will go so comparatively smoothly. Especially as the next disaster is already looming on the horizon for many VMware customers: VMware set-ups are often used in conjunction with NAS appliances for iSCSI. However, NetApp, Dell-EMC and various other providers have also continuously tightened the price screw in recent years. Anyone who purchased their last turnkey NAS five years ago will probably soon be in for the next unpleasant surprise. This would have been both foreseeable and avoidable if free standards had been adhered to.
(mack)
