Posts made by dyasny

@FATeknollogee the install itself has been covered too many times to mention Maybe if he runs into a specific issue and solves it, that would be a good thing to share

@FATeknollogee there isn't that much to setting oVirt up, it's not a new untested pile of code you have to work hard to get from github and build, it's a complete product, pretty easy to set up and use. Initially it was meant as a VDI system for Windows admins, so it was made very easy to set up from the get-go.

@scottalanmiller said in Replication Options for KVM to DR Site:

It's not a business tool. It's officially unsupported and puts your VMware installation at risk. If you use Zerto, you dont' believe in VMware. It is strongly advised officially to never, ever consider.

If you use it, you voluntarily suspend any support guarantees from everyone. It breaches every possible support position. Not that Vmware would use that as an excuse to not support you, VMware isn't like that, but any legal responsibility that you've paid for them to have is completely gone.

Is that an official documented vmware policy? I wouldn't be surprised of course, they are horrible at actually being able to cope with competition, hence all the restrictions on benchmarking and comparisons.

I'm not a fan of Zerto (for my own reasons, I wouldn't go near them, ever), but if vmware are basically selling you a product and then refuse to support it (even if support has been paid for) because you use an ISV for added functionality instead of paying for their own add-on, that's just ugly of them. Another reason to run away to other vendors, IMO.

@scottalanmiller said in Replication Options for KVM to DR Site:

That's my point. Losing the backups as well. Backups still don't protect from everything. Not when you lose whole regions. There are events that take out your backups as well, even if they are stored in another country, for example.

OK, lets go to the extremes, the planet gets hit by a dinosaur killing meteorite, we all go into a new ice age and die. DR becomes a non-issue. I thought I already mentioned the level-of-paranoia variable. IF you want to protect yourself from such an event - invest in sending backups to Mars and storing them there. But really, I thought we were having a meaningful conversation here, not a post apocalyptic scifi con

@scottalanmiller you're obviously not inventing the term HA, but you are inventing this ridiculous saying about HA being what you do and not what you buy. You can, of course, hack HA into almost any service, but a product that is already built with HA in mind is something you buy and use as designed - and you get HA. Out of the box, if you bought and configured all the prerequisites. oVirt, vCenter, and a ton of other products have it designed into them, so if you pay for it, and for the hardware that supports it, you can have it right there out of the box, if you follow the setup guide. Everything else is just you throwing meaningless pronouncements in the air.

You buy VMware, with the HA features (don't remember if those cost extra, doesn't matter here). You buy hardware that supports whatever VMWare uses for HA (IPMI/redfish/redundant switches etc - whatever is the best practice) and you follow the config guide to set it up - you have yourself highly available VMs, with all the standard properties for HA - downtime SLA, splitbrain avoidance etc etc. These are features you pay for (that's what "buy" means in the English language), both on the software and hardware side of things.

And yes, I've decided arguing with you here is a huge waste of time, because for every comment you come back with 10, and I have no bandwidth for replying to that much, so if you think you "won" an argument or whatever tickles your fancy, sure, go ahead. I'll just answer if I want to, at my own convenience. Hope you don't mind.

@scottalanmiller I've worked for a few banks as well, nothing to be too proud of Seriously though, when you recover it is up to you to build the infra and get everything ready for recovery, when you avoid you need to target specific issues you're avoiding. You can restore from backup in pretty much any reasonable scenario. You can't avoid every, even reasonable, scenario, this is why backups were invented.

@scottalanmiller said in New Infrastructure to Replace Scale Cluster:

By definition, HA can't be provided "out of the box." HA is something you do, not something you buy. A product may have features to make HA easier, but a product itself can't do HA.

In an IPOD, oVirt would simply automate LA (low availability). HA must be significantly higher than standard availability. The proposed IPOD design results in significantly lower than standard. (Where standard is an enterprise server with local storage and no system of this kind whatsoever.)

This is just a bunch of terms you invented on the spot. No truth to them. HA can be provided "out of the box" by a system that is capable of it. It is not "something you do", it's a product or system feature.

@scottalanmiller try an ec2 i3.metal

@scottalanmiller said in New Infrastructure to Replace Scale Cluster:

@dyasny said in New Infrastructure to Replace Scale Cluster:

I've never seen a well built SAN go completely down in over 20 years of working with them.

I have. Most SANs fail with reckless abandon. Really good ones are incredibly stable, but everything fails sometimes.

Now, in the storage industry, a "good SAN" would be defined as one that is a part of a cluster. No single box is every that reliable, even the best ones are easily subject to the forklift problem if nothing else.

SANs can approach mainframes in reliability, but to do so is so costly that no one does it. In the real world, that kind of storage carries really high risks and/or cost compared to other options.

let me rephrase myself. I've seen disks, controllers, PSUs, even backplanes and mobos fail in SANs, None of that ever caused an actual outage.

@scottalanmiller :drum:

@scottalanmiller said in Replication Options for KVM to DR Site:

@dyasny said in Replication Options for KVM to DR Site:

@scottalanmiller said in Replication Options for KVM to DR Site:

@dyasny said in Replication Options for KVM to DR Site:

that's because DRBD isn't a DR tool, it's a block replication tool. All real DR solutions have granularity and versioning. SRM, Zerto, NeverFail, even old WanSync - all of them could do it. DRBD can only sync blocks, nothing else. This is not DR.

Like everything in the UNIX world, you build solutions from individual blocks. In the Windows world, people assume all solutions will do a single task and be pre-assembled for them. So you find a "product" that does what you need.

In the UNIX world, it is generally assumed that you'll understand the tools and the concepts and use the existing tools transparently and put together what you need it to do.

So nothing is a DR tool in UNIX, but DRBD is a perfectly valid building block to handle one of the pieces that becomes a DR solution.

To paraphrase John Nicholson of VMware... disaster recovery is something that you do, not something that you buy.

Well, no. DRBD (or any replication or storage sharding really) can be used if you want to protect yourself from a storage failure, just like with RAID you protect yourself from a disk failure. A DR solution should be able to bring you back from an outage. There is a huge difference there. Tape backup is more of a DR solution than RAID or storage replication. That's the point I'm trying to make.

Yes, we use the term DA vs DR.

RAID (including Network RAID like DRBD) is Disaster Avoidance, backups are part of Disaster Recovery.

One is to prevent a disaster, the other is to recovery from a disaster.

Exactly. This is why DRBD is not a DR solution, like I've been saying all along. The problem with any kind of avoidance is that it can only target a limited number of potential problems, while a recovery solution covers everything.

@scottalanmiller said in Replication Options for KVM to DR Site:

@dyasny said in Replication Options for KVM to DR Site:

that's because DRBD isn't a DR tool, it's a block replication tool. All real DR solutions have granularity and versioning. SRM, Zerto, NeverFail, even old WanSync - all of them could do it. DRBD can only sync blocks, nothing else. This is not DR.

Like everything in the UNIX world, you build solutions from individual blocks. In the Windows world, people assume all solutions will do a single task and be pre-assembled for them. So you find a "product" that does what you need.

In the UNIX world, it is generally assumed that you'll understand the tools and the concepts and use the existing tools transparently and put together what you need it to do.

So nothing is a DR tool in UNIX, but DRBD is a perfectly valid building block to handle one of the pieces that becomes a DR solution.

To paraphrase John Nicholson of VMware... disaster recovery is something that you do, not something that you buy.

Well, no. DRBD (or any replication or storage sharding really) can be used if you want to protect yourself from a storage failure, just like with RAID you protect yourself from a disk failure. A DR solution should be able to bring you back from an outage. There is a huge difference there. Tape backup is more of a DR solution than RAID or storage replication. That's the point I'm trying to make.

@scottalanmiller said in Replication Options for KVM to DR Site:

It's Network RAID. No RAID has versioning. If you want versioning, which is a great thing to have, that happens at a higher level. You can still do it, it's just not included in DRBD.

Exactly! Saying RAID is DR, is like saying it's the same as backup - only noobs do that.

@scottalanmiller said in Ovirt:

I'm doing an oVirt install right now.

Cool hope you have it planned properly

@scottalanmiller LOL

@scottalanmiller said in Testing oVirt...:

@dyasny said in Testing oVirt...:

Why DRDOS? NWDOS-7 was great, it could even do multithreading and networking

NWDos was just a rebranding. Still DR-DOS.

https://en.wikipedia.org/wiki/DR-DOS#Novell_DOS

oh I know, but Novell added the networking stack in there, and iirc the multitasking was also developed in cooperation with Novell

@JaredBusch

Would never use, ever. Hacked ass shit.

I'm not a fan either, but it is a solution available out there

@JaredBusch said in Replication Options for KVM to DR Site:

You are not understanding what DR is.

Looks like I could say the same thing to you.

DR has no mandate for versioning. Merely a means of recovering from a disaster.

DR is for disaster recovery, exactly. An encoded disk can be as much of a disaster as a downed DC. Only DRBD would replicate the encoding, irrecoverably. DR solutions are there to bring you back from being down. Both replication based and copy based (backup) DR solutions have a PIT factor, otherwise they are not DR solutions, but merely a means of replicating a bunch of blocks. Have you ever seen Netapp advertise their metroclusters as DR solutions for example?

When it comes to any solution that does replication, that is not, ever, supposed to be a point in time recovery. Replication DR is for site failure. Nothing more and nothing less.

You are plain out wrong here. Replication DR is for site failure. And it is NOT an HA solution, it is not meant to perform a failover. It is there so that if the main site fails, you can bring up a secondary site within the configured SLA. That SLA always includes some work time lost. And that lost time always includes several PIT granularity periods.

@scottalanmiller said in Testing oVirt...:

@dyasny said in Testing oVirt...:

NT is what killed it.

Yeah, Netware on DR-DOS was very.... aged by that point.

Why DRDOS? NWDOS-7 was great, it could even do multithreading and networking

@scottalanmiller said in Testing oVirt...:

@dyasny said in Testing oVirt...:

Novell used to be bigger than Microsoft

That's been a very long time 1992 probably. Pre-NT era.

NT is what killed it. But in the mid-90s I had netware 3 based clusters serving multi-site locations with frame-relay links, running IPX/SPX and diskless clients. Good times