VMware hits back at Amazon cloud Trojan Horse with ... a blog post VMware has responded to a Trojan Horse bit of tech from Amazon with a blog post disparaging the rival's approach. Last Friday, Bezos & Co. announced the "AWS Connector for vCenter" plug-in, which lets admins buy, manage, and migrate Amazon Web Services cloud VMs from within the familiar vCenter admin environment. This free …
Jack, buddy, I have huge respect for you...but I feel I must disagree with your assessment of the blog post being aught but marketing malarkey. The fellow behind that blog post raises some damned good points, especially around bringing workloads back into your on-prem datacenter once you're done.
Cloud bursting with VMware/vCHS isn't great, but it's a heck of a lot better than this Amazon connector allows. Beyond that, to be blunt, VMware has some great next-gen technologies in QA around vCHS that will make cloudbursting easier. I'm sure you've seen the same NDAed slides we all have at this point; it's all an open secret by now.
All of which leads me to: Amazon's move is the desperate one. Bullshitting saved for another day, Microsoft has the best hybrid cloud. This is followed by both companies that have deployed Openstack internally; there are hundreds of Openstack public providers and Openstack to Openstack actually works quite well.
VMware is next up, their technology is immature, but they are dumping amazing resources into it. The people working on the hybrid cloud offering at VMware are some of the brightest on the planet and I promise you they will be at an Azure level by the end of the year. They'll probably pull away from MS and have the best damned hybrid offering (at the highest price!) of all contenders by VMworld 2015.
All of which leaves Amazon, where? As the poster child for voluntarily handing your data to the US government? The embodiment of the inability to even attempt data sovereignty or control over your own workloads in a superficial way? Amazon is great for SaaS developers who make pointless tat or who work in industries where America basically sets global law anyways. (See: Netflix.) It's rather less awesome for the man - many - high-value industries that are either regulated, or where innovation occurs at a such a pace that economic/industrial espionage* is something that companies worry about.
The public cloud isn't safe for some workloads. On-premises isn't cost effective or fast enough for other workloads. That makes hybrids cloud an absolute necessity and it is Amazon - not VMware - that doesn't have a story here.
There are some very valid concerns about picking up your workloads and putting them on a public cloud, regardless of which cloud you choose. But when the workloads can't come back easily, or your VMs are converted, or you are integrating with management tools/using software with weird licensing restrictions then things get a hell of a lot more messy than "this is technically possible."
We could always take our workloads and put them into Amazon's cloud. The thing that was holding us back was never an integration tool. It was all the myriad reasons listed in that blog post, and more besides.
When Amazon develops the ability to truly move workloads from on-prem to the cloud and back again, with conversion headaches, networking issues and management/agent integration tools dealt with on the fly, then VMware should start sweating. Until then, I'm pretty sure that VMware's best path forward is to make a dmaned good hybrid solution of their own...and lower the prices for service providers dramatically.
If they don't, Microsoft is going to win. Microsoft has a hybrid cloud that is not just on-prem and public cloud, it's "service provider"...and that's critical. Data sovereignty means a lot of people want cloudbursting...but only within their own legal jurisdiction. Microsoft has an answer to this. VMware doesn't**.
VMware has about a year, maybe a year and a half to get that sorted before even large enterprises are willing to use the abomination that is SCVMM*** in exchange for a proper multi-teir cloud.
*Ask yourself: if you had the cure for cancer, the formula for room temperature superconducters or the plans for a machine that could cancel gravity in a localized field would you store that information with an American cloud provider? If you would, please e-mail me and we can discuss a fantastic opportunity I have regarding some riverside real estate that provides access for individuals wishing to cross.
**Because most cloud providers won't pay VMware's exorbitant fees and are still miffed that VMware is competing against them.
***Fuck SCVMM.
"There is no easy way to move workloads back to one of your data centers[..]"
I remember when I was a kid living in Hawaii for a bit there would be these roach motels(they actually had that as their name on the device) under furniture sometimes, little things where roaches would go in and get stuck and die. You check in but you don't check out..
amazon is a roach motel. The poor service offerings just make the whole situation even worse.
Last night I did have my first vmware host crash... in **8 years**. The system PSOD'd and the HP firmware auto rebooted the box. Still waiting on support to analyze the logs first indication is it is a General protection fault (13).
By the time our alerting went off vmware HA had already restarted all of the VMs on other systems(though there was still some manual recovery steps needed on a few systems).
Compared to the few years I spent in amazon, oh my what a nice experience vmware is/was.
Oh and "bursting" to a cloud is a fantasy for the vast vast vast majority of organizations out there, most applications simply don't scale that way(and won't because that takes hard work and most orgs rather add other features).
"he system PSOD'd and the HP firmware auto rebooted the box. Still waiting on support to analyze the logs first indication is it is a General protection fault (13)."
$20 it's a bad SPD chip on a single DIMM. Virtually impossible to isolate. The error manifests as a PSOD/BSOD only under certain very specific conditions. The issue is that the BIOS is configured to clock each DIMM (or at least each bank) independently. The bad SPD chip reports an incorrect speed for the capabilities of the DIMM. The result: an overclocked DIMM that goes squirrely seemingly at random, but especially when the temperature goes up.
The error will often show up as a set of ECC errors within your system, but when you go to memtest the DIMMs individually they're all fine. Alternately, you could have a system wherein timings are set per bank, not per DIMM, and the bad SPD chip is in fact on a DIMM that absolutely can handle the higher speed, but one of the other DIMMs in the bank can't, and that (perfectly fine, if tested on it's own) DIMM is the one that errors out.
Solution: attempt to isolate DIMM (hard) and RMA - or - manually set the speed and timings of all DIMMs in the system. (I typically downclock to just below rated anyways, just to avoid minor manufacturing defect issues.)
You can also use that Intel technology that allows you to RAID 1 your RAM. For file servers, this is what I do: RAID 1 the RAM and downclock it. Then the things run like a tank.
As I have said many times on the Register. VMware is expensive (at the moment) but very very smart. It is robust, guaranteed and game changing, and it will not stop there. 90% of Enterprise Hyper-V deployments break and require MS support intervention = FUDware. Azure is much better and MS is going to kill of the MCSE and MCITPs that supported and worshipped them. So ironic!
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
