I think I am missing something about Hyper-V....?
-
@obsolesce said in I think I am missing something about Hyper-V....?:
@phlipelder said in I think I am missing something about Hyper-V....?:
Cosmos Darwin: Storage Spaces Direct: 10,000 Clusters and counting
I'm curious to see how much permanent data loss has occurred as a result of using S2D.
Heh … I've seen many a disaster recovery situation where backups were either non-existent or failed with no indication of things being broken. We've been in them (BUE ugh) and have been called in on many more. If the backups have not been fully restored bare-metal or bare-hypervisor they are not proven thus no-good IMNSHO. Spot restores do not count.
Here's a bad one: Maxta
Slightly different situation but: Dell Stops Swonly Sales of ScaleIO Virtual SAN <-- Folks being left in a lurch.
Data loss responsibility starts with the folks managing the backups. Usually, they are not important and seen as a cumbersome expense until something goes blotto. Then all of a sudden the CxOs are throwing money at a proper disaster recovery setup. But, by then it's too late.
-
@phlipelder said in I think I am missing something about Hyper-V....?:
VMware has had some spectacular bugs with one of the latest brought to light by Veeam with data loss a very real possibility.
No software product out there is perfect. That does not excuse the early release cycles that we are seeing from many vendors not just Microsoft.I"m not aware of Veeam identifying any corruption for primary (VMFS/NFS/vSAN/vVols) storage.
Comparing a niche bug on VADP under specific situations could cause a differential backup to be corrupt to one that corrupts primary storage is...
Like comparing a flat spare tire, to the problem with the Cobalt that causes it to disable the airbags and crash. People have far higher expectations for primary storage, and they expect clear communication.Note, Veeam (and VADP based backups) are relatively easy to test with SureBackup (how Veeam found the last issue that I'm aware of).
-
@phlipelder said in I think I am missing something about Hyper-V....?:
Slightly different situation but: Dell Stops Swonly Sales of ScaleIO Virtual SAN <-- Folks being left in a lurch.
Dell's server division (who took over ScaleIO ownership from CPSD) didn't want to sell it on HPE servers. That shouldn't really shock anyone.
-
@phlipelder said in I think I am missing something about Hyper-V....?:
@scottalanmiller said in I think I am missing something about Hyper-V....?:
@storageninja said in I think I am missing something about Hyper-V....?:
@phlipelder said in I think I am missing something about Hyper-V....?:
Why Hyper-V? We don't have to pay anything extra to build-out a virtualization platform for one.
We've done well with Hyper-V and Storage Spaces.Storage Spaces Direct is Datacenter licensing only. If you have core dense platforms this gets expensive.
And it isn't production ready, and doesn't have production readiness on its roadmap. And that's right from the MVPs. It's a joke that MS released way too early with no way to get working. In the enterprise space, it's essentially non-existent and those that have used it have been burned big time.
Microsoft knows what's happening with their products and how they are being used.
Cosmos Darwin: Storage Spaces Direct: 10,000 Clusters and counting
There are deployed to production S2D clusters out there. And, like everything else out there, there's always first-run jitters and issues.
We've seen issues in all Windows Server versions out of the box since 2008 R2 RTM and even earlier.
VMware has had some spectacular bugs with one of the latest brought to light by Veeam with data loss a very real possibility.
No software product out there is perfect. That does not excuse the early release cycles that we are seeing from many vendors not just Microsoft.
As far as licensing S2D goes, we SPLA the DC license with our SMB deployments starting at 10-15 seats and up. They are also great ReFS repositories for Veeam (something they request to have under their backups).
Yeah, 10K clusters, but they aren't production and they aren't working. Insider info says S2D is dead and no serious use cases of it are out there. MS knows VERY well that it's not being adopted.
With companies of this size, it's trivial to make marketing data about cluster counts. 10K isn't even the number you'd expect in early labs.
We aren't talking early jitters, we are years past that, this is MS core MVP user base and enterprise customers finding it unusable.
-
@obsolesce said in I think I am missing something about Hyper-V....?:
@phlipelder said in I think I am missing something about Hyper-V....?:
Cosmos Darwin: Storage Spaces Direct: 10,000 Clusters and counting
I'm curious to see how much permanent data loss has occurred as a result of using S2D.
Very little, because no one is using it in production. This helps them hide just how bad the situation is. Only 10K clusters means it's not even being tested heavily.
-
Maybe Hyper-V is more well known by more sys admins world wide and that's why it's so widely used?
Everyone knows Windows Server, it's a natural and free progression from there.
-
@siringo said in I think I am missing something about Hyper-V....?:
Maybe Hyper-V is more well known by more sys admins world wide and that's why it's so widely used?
Absolutely, marketing, not tech or business utility, drives 99% of IT decisions. As @StorageNinja once said, 99% of Hyper-V installations are done by accident or by mistake. People think it's supported when they install it, or they think it is requires for Windows or they think it gives them special licensing. Hyper-V is an okay product, but ask people who have been deploying it for any length of time, and nearly all of them will give you a reason that they chose it that is false, and almost always it's the same tired licensing mistake that tells us they don't know literally the first thing that everyone learns about Hyper-V.
Ask: What's the first thing to know about Hyper-V.
And everyone knows that it is "It's free and has no license and gives no licensing benefits."
Yet 90% of people using Hyper-V openly state that they use it because they "go the license as part of a package" or that they "need it because of Windows licensing", both of which contradict the very first thing you talk about with Hyper-V! It means the average deployment has never had even the most casual conversation about or investigation. It was not understood and someone deployed it thinking that it was something that it wasn't, and they kept it either because they never figured it out, it was too late, or whatever.
-
@siringo said in I think I am missing something about Hyper-V....?:
Everyone knows Windows Server, it's a natural and free progression from there.
In the world of IT doing things based on "what they heard of" and not "doing their due diligence", absolutely. There is no surprise at all that it is everywhere in bad installations.
Not that good installations don't do it, it's just that they are not the norm.
-
@scottalanmiller said in I think I am missing something about Hyper-V....?:
As @StorageNinja once said, 99% of Hyper-V installations are done by accident or by mistake.
John White and I used to keep a list of people who were mistaken that Hyper-V had special licensing rights. At one point it was over 75% of why people went Hyper-V.
-
@storageninja said in I think I am missing something about Hyper-V....?:
@scottalanmiller said in I think I am missing something about Hyper-V....?:
As @StorageNinja once said, 99% of Hyper-V installations are done by accident or by mistake.
John White and I used to keep a list of people who were mistaken that Hyper-V had special licensing rights. At one point it was over 75% of why people went Hyper-V.
And, this is huge, that's 75% that had to have admitted it.... which is astounding given that it must have been almost always in a venue where people were constantly pointing out how silly that myth was. So anyone admitting it had to be very willing to look quite foolish. It's nuts that 75% even fessed up to that mistake under the conditions! The actual number has to be way higher.
-
@siringo said in I think I am missing something about Hyper-V....?:
Maybe Hyper-V is more well known by more sys admins world wide and that's why it's so widely used?
Everyone knows Windows Server, it's a natural and free progression from there.People who's entry into the field is Microsoft certification training this may be true, but I find it's just enough to deploy and troubleshoot a non-clustered instance. Once you mix in clustering their ability to deal with CSV's generally isn't there.
Some small cluster deployments are because Microsoft paid (quite a lot of money actually) to the partners to stand up a Hyper-V Cluster. Throwing money at the channel will get a product deployed until it burns someone.
-
@storageninja said in I think I am missing something about Hyper-V....?:
@siringo said in I think I am missing something about Hyper-V....?:
Maybe Hyper-V is more well known by more sys admins world wide and that's why it's so widely used?
Everyone knows Windows Server, it's a natural and free progression from there.People who's entry into the field is Microsoft certification training this may be true, but I find it's just enough to deploy and troubleshoot a non-clustered instance. Once you mix in clustering their ability to deal with CSV's generally isn't there.
I think that he means that the name is more well known. I think it's very obvious that the product itself is not well known at all, and especially not known or understood by Windows admins. They are specifically the group that seems to have consistently no base level knowledge of it, it's relationship to Windows, and constantly repeat the "everyone has been told" myths over and over.
If you talk to other IT disciplines, knowledge of Hyper-V seems "normal". But specifically in the Windows Admin ranks, the pervasive misinformation appears to have taken such a firm root that the group you'd expect to have the best passing knowledge of it seems ot have the least.
-
@storageninja said in I think I am missing something about Hyper-V....?:
Some small cluster deployments are because Microsoft paid (quite a lot of money actually) to the partners to stand up a Hyper-V Cluster. Throwing money at the channel will get a product deployed until it burns someone.
This is what explains the numbers around things like S2D, too. Channel pushing it where it makes no sense and the customers are too small so that when it fails, no one hears about it.
-
@storageninja said in I think I am missing something about Hyper-V....?:
@scottalanmiller said in I think I am missing something about Hyper-V....?:
As @StorageNinja once said, 99% of Hyper-V installations are done by accident or by mistake.
John White and I used to keep a list of people who were mistaken that Hyper-V had special licensing rights. At one point it was over 75% of why people went Hyper-V.
What do you mean by this... that those people thought Hyper-V gave them special licensing rights to running Windows guests or something?
-
@obsolesce said in I think I am missing something about Hyper-V....?:
@storageninja said in I think I am missing something about Hyper-V....?:
@scottalanmiller said in I think I am missing something about Hyper-V....?:
As @StorageNinja once said, 99% of Hyper-V installations are done by accident or by mistake.
John White and I used to keep a list of people who were mistaken that Hyper-V had special licensing rights. At one point it was over 75% of why people went Hyper-V.
What do you mean by this... that those people thought Hyper-V gave them special licensing rights to running Windows guests or something?
Correct. People who deployed Hyper-V because they believed it provided something that was not true.
-
@scottalanmiller said in I think I am missing something about Hyper-V....?:
@obsolesce said in I think I am missing something about Hyper-V....?:
@phlipelder said in I think I am missing something about Hyper-V....?:
Cosmos Darwin: Storage Spaces Direct: 10,000 Clusters and counting
I'm curious to see how much permanent data loss has occurred as a result of using S2D.
Very little, because no one is using it in production. This helps them hide just how bad the situation is. Only 10K clusters means it's not even being tested heavily.
"No one is using it in production." <-- False.
Making broad brush stroke claims like that is fallacious at best.
There are in-production S2D clusters.
-
@phlipelder said in I think I am missing something about Hyper-V....?:
@scottalanmiller said in I think I am missing something about Hyper-V....?:
@obsolesce said in I think I am missing something about Hyper-V....?:
@phlipelder said in I think I am missing something about Hyper-V....?:
Cosmos Darwin: Storage Spaces Direct: 10,000 Clusters and counting
I'm curious to see how much permanent data loss has occurred as a result of using S2D.
Very little, because no one is using it in production. This helps them hide just how bad the situation is. Only 10K clusters means it's not even being tested heavily.
"No one is using it in production." <-- False.
Making broad brush stroke claims like that is fallacious at best.
There are in-production S2D clusters.
The point being, extremely few. To the point of being statistical noise.
-
@phlipelder said in I think I am missing something about Hyper-V....?:
"No one is using it in production." <-- False.
figure cited is the number of currently active clusters reporting anonymized census-level telemetry, excluding internal Microsoft deployments and those that are obviously not production, such as clusters that exist for less than 7 days (e.g. demo environments) or single-node Azure Stack Development Kits. Clusters which cannot or do not report telemetry are also not included.
When you have 10K clusters, and your competitors have (Singular) customers closing in on that many clusters that's noise. VMware is reporting over 14K Customers.
The industry standard for reporting product usage is reporting total customer count. Microsoft either knows is embarrassing low, or outright doesn't have the information. My lab which includes a SSD cluster (which is nested on top of vSAN) qualifies as "production" by this definition. Hell if Starwind or Veeams QA lab is leaving stuff on over 7 days he's a production customer!
For comparison VMware and Nutanix both state the number of paying customers they have on earnings calls. This is audited, and making a false statement here could incur jail time. Since both sell discrete software SKU's for their HCI product (and can track generation of licensing keys in the portal) they can give real and actually specific numbers. VMware's phone home tracking (CEIP) can at least tell if a licensing key is a production key vs. a demo or lab license key. It also can tell if the deployment is nested (is the hardware detected as "VMware or some sort of paravirtual device) making it easy to filter and identify. Microsoft including all clusters more than 7 days old that were not on their network and include non-licensed clusters is highly dubious.
When a vendor uses odd qualifications to report customer adoption that are non-standard and use vague criteria you assume the reality doesn't smell like roses.
-
@phlipelder said in I think I am missing something about Hyper-V....?:
As far as licensing S2D goes, we SPLA the DC license with our SMB deployments starting at 10-15 seats and up. They are also great ReFS repositories for Veeam (something they request to have under their backups).
This only works if they are on your servers, running in your datacenter. You can't use SPLA for customer owned gear.
-
@storageninja said in I think I am missing something about Hyper-V....?:
This only works if they are on your servers, running in your datacenter. You can't use SPLA for customer owned gear.
Last I checked, you can use SPLA on customer gear. Rule was changed about 1-2 years ago, IIRC.
