Home > VMware, VMworld > VMworld: Juggling and jousting, VMware’s battles for the future

VMworld: Juggling and jousting, VMware’s battles for the future

August 28th, 2015

VMware’s annual love fest that is VMworld kicks off in a few days time. VMworld isn’t just any vendor conference, it has become THE central meeting place to network and trade. Due to VMware technology being ubiquitous at the heart of the data center, nearly every vendor selling to the SMB, government or enterprise partners or interoperates with VMware meaning attendees can not only gorge themselves on VMware goodies but meet and catch-up with all the other vendors they also use, a very useful one-stop-shop conference.
VMware’s community has been the social gathering ground for much of the industry so getting up to date with technology is only half the fun, catching up with Twitter friends and meeting people from across the world is often the real reason people attend. Speaking to peers often teaches you more about the real world of IT compared to any amount of vendor marketing.

The world of IT doesn’t hang around, there’s constant change in how we do things and VMware is a prime target for change as everyone wants a piece of their pie. Nowadays there are options, VMware’s hypervisor isn’t the only viable one any more, Microsoft’s Hyper-V is good enough but without the bells and whistles and KVM, based on code you can change yourself is being rejigged and added to by many companies eager to fill in its shortcomings. Speaking of bells and whistles and shortcomings, these are the things that VMware’s hypervisor excel at which has made them rich but the bells and whisles are now no longer mandatory. New apps don’t need to rely on available infrastructure, you dont HAVE to have vMotion.

VMware is seen as rich pickings on so many fronts it must feel it is fighting battles in every corner it operates but that’s a sign of healthy competition which ultimately benefits customers. The fight is not just on the private infrastructure front but every type of cloud possible and also new ways of application development. There’s a lot going on in the IT world and much to infer by what is said or even not said at VMworld!

The People Juggling

VMware itself is changing. looking slightly further up to parent company, EMC, there’s shareholder sabre rattling at the feet of EMC super boss, Joe Tucci, with plenty of rumours of who may be buying whom. Ex VMware CEO Paul Maritz who left to run Pivotal still within the EMC club has sort of retired, does this mean VMware CEO Pat Gelsinger with bags of heavy technical experience at Intel is destined for other things? What else is on the horizon for some more deck shuffling within the federation club house?

VMware’s newish chief CTO (there are a few across the company), Ben Fathi, unexpectedly left just before VMworld without saying where’s he’s going so was he pushed or did he jump? Ray O’Farrell with deep technical chops in R&D has now been appointed CTO. It remains to be seen whether O’Farrell can add chief evangelist to his chief technologist street cred (I’m sure the Irish charm can’t help!). Ex CTO, Steve Herrod, arguably the standout evangelist+technologist VMware has ever had who left to the venture capital world is being brought back into VMworld to keynote their developer relations DevOps days so has not gone far away, will he be re-joining in some capacity…CEO perhaps with Pat Gelsinger moving, perhaps taking over from Paul Maritz or even Joe Tucci, due to retire sometime. Pat must be a demanding boss for a CTO as he’s deeply technical and strategic himself.

VMware rightly has been building up its own ecosystem so it isn’t reliant on just hypervisor revenue. The Software Defined Data Center term incidentally from the division where new CTO Ray O’Farrell comes from is about removing custom hardware infrastructure from your data center and running everything in software, preferably VMware software of course but running on cheaper generic hardware with software revenue going to VMware rather than a hardware vendor.

VMware has two major areas of attack for this, storage and networking.

Storage Jousting

Storage is all about VSAN, VMware’s home grown storage offering which runs on local disks on standard server hardware without any requirement for an external shared SAN. VSAN hasn’t been out for long, it is now on version two baked into the vSphere 6 hypervisor. Remarkably as a new storage offering, VSAN is doing pretty well and actually being adopted. Timing has been on VMware’s side with the rise of flash performance available within servers. VSAN is still missing some basic stuff like de-duplication and native replication but these must be on the VSAN team roadmap.

Chuck Hollis, an EMC veteran who moved to VMware two years ago to champion VSAN as chief evangelist, is suddenly and surprisingly heading to Oracle to join a previous boss to promote Oracle’s generally unknown converged infrastructure. Chuck has been a very vocal marketer schooled in the hard knocks of the storage business where aggressive marketing is the norm. Some have bristled at the way he’s vilified anyone not using VSAN, setting his sights particularly on storage/hyper-converged competitor Nutanix with claims of subpar performance, excessive resource use, higher cost and more complex management. There is unlikely any actual truth in either side of this spat with marketing wins and fails on both sides, just different opinion and choices so the posturing from VMware has been seen by the community as perhaps unnecessary and a relic of the tactics EMC is famous for. Chuck may have rattled the cages but was he effective? Perhaps the fallout will see people challenging their storage vendors more and thinking there must be some substance if there’s a fight and trying out VSAN?

VMware finds itself in a difficult position. It is obliged to join the standard storage industry mudsliging to compete and push VSAN as hard as it can but yet VMware partners with every other storage vendor so has to play nice as well. A juggling act which is very difficult. Duncan Epping, the virtualisation world’s No. 1 blogger and long time VMware evangelist is taking on Chuck’s role so it will be interesting to see how he navigates this seemingly conflicting path. Duncan is highly respected within the community for his technical ability and generally calm public demeanor but he’s been part of VSAN from its very beginning so may be seeing his VSAN baby as the only answer to any question regardless of what other vendors may offer. VMware is also struggling with its hyper-converged offering EVO:RAIL which Duncan was instrumental in designing which is basically VSAN and some new management software on standard partner hardware. Not all partners are keen as they lose juicy software revenue in exchange for lowly service contracts on low margin hardware. HP has pulled out as they have their own VSAN-like software, StoreVirtual. EVO:RAIL doesn’t integrate with existing VMware environments so it means you need to build a new segregated environment, great for remote offices but requiring 4 nodes minimum and some ridiculous pricing has made it a tough sell, so much for championing commodity hardware which should be cheaper yet then selling eye-wateringly expensive software. EVO:RAIL needs a big price cut and more connectivity to other vCenters to gain traction. Virtual Volumes is meant to bring VSAN’s policy based management to external arrays and has been years in the making and will still take years to get mainstream adoption as policy based management doesn’t yet fit into how enterprises provision external storage. Hyper-convergence in the meantime will have far more of an impact, with or without policy based management.

Network Jousting

Next is networking, VMware see it’s software defined network product, NSX, bought from Nicira, as hopefully the next big thing to hit infrastructure and is dying to be the undisputed future data center network leader so it can shift income from ESXi to NSX. Unfortunately, it isn’t 2003 and this will be one of it’s bigger challenges. Virtualisation with ESX was such a game changer as it saved companies money on day one and didn’t involve any other infrastructure change, just squeeze a few more virtual servers without changing the OS on a physical server and the rest was the same. Networking isn’t like this as it’s all about connecting different things together across the world. Introducing any new networking technology is a very very long battle and this time they have a challenger in Cisco with big pockets, big muscle and their own massive ecosystem. Cisco’s own SDN product, ACI, could very easily win over NSX just because most companies already use Cisco and see no reason to change. In order to adopt NSX you really are looking at ripping out your existing switches, companies only change their network infrastructure at least every 5 years if not longer and with NSX being unproven compared to Cisco with or without ACI, this is going to be a very very long battle. NSX may have more luck with service providers looking for more flexibility but these aren’t high paying customers compared to enterprises.
NSX 6.2 was recently released which didn’t seem groundbreaking, it added support for multiple vCenters which now means you can span NSX networks across data centers. Previously NSX was only suited to a single data center, not very cloudy. 6.2 fixes numerous bugs and still hasn’t worked out the kinks in many others so confidence isn’t high yet. NSX has a very very long way to go just at the time when companies are moving things to public clouds like Amazon and Azure where NSX is obviously nowhere to be seen, the potential market is shrinking. Expect to hear a lot about NSX and cloud mobility at VMworld.
EVO:RACK is EVO:RAIL’s bigger sibling which will incorporate NSX and vRealize cloud management software to build a pure VMware software Vblock/FlexPod alternative. It’s going to be a very difficult sell with unproven NSX, complicated cloud management and a drive towards hyper-converged. I’m sure we will hear more about EVO:RACK at VMworld.

Cloud Jousting and Juggling

VMware is spending a vast amount of time, effort and cost to build out its public cloud vCloud Air but how relevant this will be in the future is very uncertain. It’s pretty much an infrastructure cloud with OK, the excellent Cloud Foundry and some other PaaS things now on top but it’s very much taking on-premises infrastructure and putting it in the cloud. VMware say this is how it should be done, you know how you run infrastructure internally, well use the same tools and processes and it will look the same on our cloud. The problem is this isn’t how people are actually moving to the public cloud. They are not migrating existing workloads and copying a 1TB VM over into a cloud and calling it a day. Enterprises are generally not moving their own workloads directly into the cloud but replacing these workloads with pure cloud offerings. That HR, Finance, training or analytics database and application isn’t being migrated, it’s being completely replaced by a SaaS offering from a 3rd party vendor with perhaps a one time data import. There’s less need to integrate apps and data on-premises with off-premises. The on-premises way of doing things is seen as old and clunky so moving the same procedure into the cloud isn’t the future.

By comparison Microsoft has built a brand new cloud platform by completely abandoning the on-premises way of doing things and built a new cloud with no legacy. Microsoft is now dragging some of this functionality back in your on-premises data center but making your own data center more cloud orientated rather than making a legacy private cloud operation in the public cloud. Would you prefer to deploy SRM and link it to vCloud Air to use storage replication to migrate an entire SQL VM to the cloud for DR or rather right-click within the SQL Management tool and select New Replica -> Cloud? As cloud moves up the stack, I feel vCloud Air is being left behind. Don’t even mention Amazon, the undisputed public cloud leader with so many API driven tools everywhere, years and years and years ahead of VMware at every level.

VMware’s promise of integrated private and public cloud management is still not a reality. vSphere and vCloud Air may use the same hypervisor but management is still actually separate other than migrations. vCenter Server is still flash based, doesn’t scale effectively and is getting more complex to deploy with each version. The rebranded vRealize Suite is meant to help with cloud management and possibly be the future but again, so much work to do. vRealize Air offerings are meant to be cloud management versions of their on-premises software equivalents but still no real integration. It is still ridiculously complex to deploy a VMware private cloud. vRealize Automation is a pig to set up, certainly no easy peasy cloud. Platform9 which is a cloud based OpenStack management product by ex-VMware employees is on the right track. Using the public cloud to manage private clouds is the way forward until you move as much of that private cloud stuff out somewhere else. People will soon need to get over the security and control issues they have, many have already with Amazon. VMware really need to drive single, simple cloud integration wherever it may be, at the moment there are too many disparate moving parts. A single place of management across vSphere, vCloud Air and vCloud Air Network is the ideal if its not too late. vCenter, vRealise Operations and vRealize Automation need to come together somehow yet be drastically simplified. People are expecting far simpler installation and operations.

Application Development

Application development is going through major upheavel at the moment. Cloud deployment is getting so easy and cheap, infrastructure is so easy to set up and platforms such as Cloud Foundry are available across multiple clouds. So called Cloud Native Applications is a newish term to refer to applications written specifically to run on clouds where the app is split apart into multiple components which are written specifically to be independently scalable within elastic clouds that don’t rely on available infrastructure.
DevOps brings the practice of developing software and operating deployed software closer together and cloud makes this so much easier. Developers can build code on their laptops and then push it to a cloud with the exact same environment and have the code run as expected. VMware is working hard to be at the forefront of these two trends. VMworld has made DevOps a central theme this year. The Cloud Native Apps team has Kit Colbert as CTO who is highly respected. Ex VMware CTO Steve Herrod is keynoting this part of VMworld so they’re throwing everything they have at it and to be with the cool kids they even have a hackathon. Interestingly much of the DevOps days is being sponsored by the vCloud Air division so a lot of DevOps at VMworld is about getting developers to get used to vCloud Air and hopefully convince them Amazon isn’t the only place to go.

VMware is also putting a massive focus on container management, the great new infrastructure deployment option that gives app components partial isolation yet shock, horror, may no longer need a hypervisor! VMware has created its own Linux distro, Photon, as a container runtime and initiated Project Bonneville which is all about developer container management at the vSphere level. Basically Bonneville makes every container its own VM but uses VMware’s clever Instant Clone feature to spin up the new VM so there’s no overhead in resource as well as deployment time and its transparent to the developer. Is this VMware forcing VMs as a solution to any problem to protect their beloved hypervisor? Containers don’t have the same isolation as a VM so VMware says it is bringing the best of both worlds to containers. By the way Microsoft will have two container runtimes. Hyper-V containers, which is what VMware is doing, which is creating a VM per container and allowing container orchestration to manage each VM as if it was a container. Microsoft will also allow native containers on Windows in the future, hypervisor optional, something VMware hasn’t said they can/will do. Will we see a world with containers running natively on ESXi one day? You bet hyper-converged rivals (Nutanix comes to mind) are building KVM based hypervisors to be able to do the same thing. With all the buzz around containers, applications still need to be rewritten so that immediately puts the brakes on widespread adoption but the signs are that the hypervisor is no longer the only way to partition applications.

There is a huge amount going on in the application development space and VMware is trying its hardest to be relevant by courting developers but will it be hypervisor and vSphere dependent or can we expect new things?

VMworld Venue Juggling

News out yesterday is that VMworld itself will be moving from San Francisco to Las Vegas next year as the Moscone Center is having some work done. People have personal preferences for each location either way which may dictate their attendance. We’ve all been used to VMworld in San Francisco with the conference and surrounding social activities embedded into San Francisco so a venue move will certainly change the show. Vegas is of course used to mega conferences and VMworld has spent time there in the past. Will VMware use the new venue to make some changes to the structure of VMworld? We will have to see.

I chose not to attend VMworld San Francisco this year as I pay my own way and thought it was time to see VMworld Europe again so I’ll be watching very closely from afar and looking forward to see how the future with VMware will develop. Have fun if you are there next week!

Categories: VMware, VMworld Tags: , ,
Comments are closed.