|
Lameish question... When we upgrade vSphere 5.5 -> 6.5 and migrate to their appliance, can we roll all the services into a single appliance or do they have to be broken up? I've been pouring through the documentation they provide and every looks like an external SSO on a Windows server migrates to an external a virtual appliance and the vCenter server migrates to a separate virtual appliance. Maybe I'm just reading the diagrams wrong, but they seem pretty drat clear about it. Currently, we have vCenter, SSO, and UM all split between three different 2K8 VM's because the original implementation needed to be as complicated as gently caress I guess. I was hoping they could all get rolled into one beefy appliance.
|
#
?
Feb 8, 2018 00:21
|
|
|
|
| # ? May 14, 2024 00:26 |
|
|
As of 6.5 the appliance is an all-in-one solution for most cases. vCenter/SSO and UM was added in with 6.5. It's the recommended path by VMware and way easier to maintain that the Windows stacks which they really half-rear end. If you need HA then you need to run 3 appliance instances: 1 primary, one standby, and one witness which chooses which is running as active. The good news is that doesn't require additional licensing like linked-mode in Windows did.
|
|
#
?
Feb 8, 2018 02:36
|
|
|
BangersInMyKnickers posted:I like RapidRecovery over Veem from technical standpoint, it gets much better compression/dedupe ratios but is IOP heavy to do it. 10k SAS arrays or high-density flash for the storage array but you can get away with a fraction of the raw disk space/spindle count compared to what Veem needs in SATA/NL-SATA. Support was iffy with the Dell acquisition and spin-off, no idea how much of a headache it is to run these days. vCenter integration made life easier for most things. I’ve never used Rapid Recovery/AppAssure, but this touches on one of my main issues with VEEAM, which is that I don’t want to have to think about the speeds and feeds of my backup targets. I also don’t want to have to think about how many proxy servers I need to create and what resources to give them and whether I want network based or San direct or some other backup type and so on. I just want a tightly integrated hardware appliance that does all of that work for me so I can just point it at clients and then point it at a long term data store for archive, and be done with it. Bonus points if, life Rubrik, I don’t even have to create jobs and schedules, just define SLAs and dump VMs in them and let the software figure out how to spread the backup load.
|
|
#
?
Feb 8, 2018 03:01
|
|
|
hatelull posted:Lameish question... I'm almost positive the rule for this is that if it's external, it will stay external during the migration. VUM will be on the vCenter appliance, regardless of how the PSC/SSO one comes over. BangersInMyKnickers posted:As of 6.5 the appliance is an all-in-one solution for most cases. vCenter/SSO and UM was added in with 6.5. It's the recommended path by VMware and way easier to maintain that the Windows stacks which they really half-rear end. If you need HA then you need to run 3 appliance instances: 1 primary, one standby, and one witness which chooses which is running as active. The good news is that doesn't require additional licensing like linked-mode in Windows did. Check the recommended topologies for vCenter 6.5 here, embedded PSC isn't the only choice (unless forced a certain way due to previous setup). Also, if you're looking at vCenter appliance HA, read through the VMware Validated Designs first (hint: in a ton of cases, the vCenter appliance HA capabilities aren't necessary, just "nice to have" for corner cases).
|
|
#
?
Feb 8, 2018 05:22
|
|
|
Tev posted:Check the recommended topologies for vCenter 6.5 here, embedded PSC isn't the only choice (unless forced a certain way due to previous setup). Also, if you're looking at vCenter appliance HA, read through the VMware Validated Designs first (hint: in a ton of cases, the vCenter appliance HA capabilities aren't necessary, just "nice to have" for corner cases). For a single VCenter environment that isn’t expected to grow beyond that there’s basically no reason to do an external PSC.
|
|
#
?
Feb 8, 2018 06:43
|
|
|
YOLOsubmarine posted:For a single VCenter environment that isn’t expected to grow beyond that there’s basically no reason to do an external PSC. Totally agreed, just making sure we know all the options available.
|
|
#
?
Feb 8, 2018 13:51
|
|
|
YOLOsubmarine posted:I’ve never used Rapid Recovery/AppAssure, but this touches on one of my main issues with VEEAM, which is that I don’t want to have to think about the speeds and feeds of my backup targets. I also don’t want to have to think about how many proxy servers I need to create and what resources to give them and whether I want network based or San direct or some other backup type and so on. RapidRecovery does a global dedupe on the entire repository (which you can keep tacking up to like 2k extents on to as you add storage), so you don't have the limitation of Veem only deduping against the data in the job which was a big complaint for me. By default you capture only incrementals on you clients on an hourly basis so you aren't scheduling backup windows over night and the amount of data transferred on those hourly jobs is very small so you don't see much impact on that end and you don't have to re-read the same blocks like you would in diff/full backup schedule. Then a nightly job runs on the server which consolidates the various snapshots down in to whatever retention window you configured (X days of hourlies, Y days of dailies, Z weeks of weeklies, etc). That got us hourly captures from all systems with a six month retention (I think my retention was 3 days of hourlies, 4 days of daily, 3 weeks of weekly, then 5 months of monthly) which mirror our previous tape rotation pretty well. That backed up about 500TB of logical data in to a 18TB 10kSAS array on a single server with no proxies. Then we had a second server off-site that we replicated to, but you can also push it to whatever cloud provider. We spec'd out something comparable on Veem with daily captures with a 6mo retention and it would have required on the order of 10x more formatted repository capacity to accomplish and at least a half-dozen servers. No thanks.
|
|
#
?
Feb 8, 2018 15:38
|
|
|
I missed my chance for auto migration to vsca when I moved* to vC 6.5 on Windows, didn't I? *On, you know, _all_ the environments
|
|
#
?
Feb 9, 2018 02:45
|
|
|
You could just make sure your hosts can be managed directly and rip vCenter down and start again 
|
|
#
?
Feb 9, 2018 02:46
|
|
|
I'm going to have to hand make a migration tool For Every God drat Linked Clone or start using Persona with all the fun that comes with a sorta roaming profile
|
|
#
?
Feb 9, 2018 02:47
|
|
|
Thanks Ants posted:You could just make sure your hosts can be managed directly and rip vCenter down and start again Horizon View adds an interesting twist.
|
|
#
?
Feb 9, 2018 02:48
|
|
|
Moey posted:Horizon View adds an interesting twist. I'm about to either become very good at horizon orchestration or at holding down liquor DevNull please please tell me you guys are working on a 6.5 Win vC to 7.0 VSCA path 
|
|
#
?
Feb 9, 2018 02:51
|
|
|
I have virtually unlimited access to 1) Push Mirage to linked clones in next update 2) Make new clones on vsca, run staggered recovery process on the day of cutover 3)  I am going to actually try this in test tomorrow, just to see what unforeseen horrors emerge in performance or what not. For starters, I need to goose sysprep... Potato Salad fucked around with this message at 02:57 on Feb 9, 2018 |
|
#
?
Feb 9, 2018 02:54
|
|
|
BangersInMyKnickers posted:RapidRecovery does a global dedupe on the entire repository (which you can keep tacking up to like 2k extents on to as you add storage), so you don't have the limitation of Veem only deduping against the data in the job which was a big complaint for me. By default you capture only incrementals on you clients on an hourly basis so you aren't scheduling backup windows over night and the amount of data transferred on those hourly jobs is very small so you don't see much impact on that end I've inherited a Rapid Recovery environment in a rough state that doesn't give the impression of being capable of scaling to the sizes you mentioned. I'm wondering if you were protecting in-guest with the agent, or VMs at the hypervisor?
|
|
#
?
Feb 9, 2018 05:35
|
|
|
Potato Salad posted:I'm about to either become very good at horizon orchestration or at holding down liquor They undoubtedly will because the windows VCenter is going away and they will have to have some way to get the last people off of it without forcing them to do hack together their own migration scripts in a repeat of the 5.x days.
|
|
#
?
Feb 9, 2018 06:23
|
|
|
Jadus posted:I've inherited a Rapid Recovery environment in a rough state that doesn't give the impression of being capable of scaling to the sizes you mentioned. I'm wondering if you were protecting in-guest with the agent, or VMs at the hypervisor? Started with agent only back before VM snapshotting existed, then moved the majority of to VMware captures with agents needing db log truncation. It's memory, cpu, and iop heavy and the nightly jobs are critical. We had someone who didn't know what they were doing run it and completely trashed all 6 months of our backups. I wouldn't move it in to production until you have a very good idea of how the guts of it work, easy to screw up if you don't understand it.
|
|
#
?
Feb 9, 2018 16:19
|
|
|
YOLOsubmarine posted:They undoubtedly will because the windows VCenter is going away and they will have to have some way to get the last people off of it without forcing them to do hack together their own migration scripts in a repeat of the 5.x days. I don't follow VC development much, but I would expect this to be the case. I can ask around today and get a better answer.
|
|
#
?
Feb 9, 2018 17:31
|
|
|
DevNull posted:I don't follow VC development much, but I would expect this to be the case. I can ask around today and get a better answer. I seriously, deeply appreciate this 
|
|
#
?
Feb 9, 2018 18:13
|
|
|
Potato Salad posted:I seriously, deeply appreciate this So what I could find, 7.0 is far enough out that there isn't even an answer to this question yet. Based on the fact that you can migrate from 5.5 and 6.0 Windows VC to 6.5 the VC appliance, I would expect them to not dead end people at 6.5
|
|
#
?
Feb 9, 2018 18:34
|
|
|
Deployed a new 6.5 ESXi host, had some weird errors, turned on SSH, then vCenter lost all connectivity to it. Going directly to the host's webpage instead of the vCenter webpage threw an "ERR_SSL_VERSION_INTERFERENCE" error, which I'd never seen before. Rebooted the host, vCenter could talk to it and I could get to the webpage. Turned SSH back on, got that error. Turned SSH off, the error went away. What the gently caress.
|
|
#
?
Feb 9, 2018 21:57
|
|
|
Is there a thread for containerization/Kubernetes stuff? Or is this the place for it?
|
|
#
?
Feb 10, 2018 20:16
|
|
|
There might be something over in https://forums.somethingawful.com/forumdisplay.php?forumid=202
|
|
#
?
Feb 10, 2018 20:25
|
|
|
enraged_camel posted:Is there a thread for containerization/Kubernetes stuff? Or is this the place for it? We kind of talk about it in the devops thread and here and the working in IT thread, but maybe it’s time for it’s own?
|
|
#
?
Feb 10, 2018 21:45
|
|
|
Maybe someone here can check my results... Setting up a new 2 host Hyper-V cluster. Each host will have a 2 processors with 8 cores each. There will be 6 Windows server VMs, and I want the ability to move them between hosts as needed. Everything I can find says I need 48 core licenses for 6 VMs of 2016 on a single host of those specifications. And if I want to run those VMs on a second host, I need to license them all over again, thus 96 cores. Am I right on this?
|
|
#
?
Feb 12, 2018 18:06
|
|
|
No. Datacenter licensing covers the core count for your physical hosts, then lets you run unlimited server VMs on that hardware. So 2 hosts, 16 cores each. 32 cores total. Datacenter comes in 16 core packs, so you just need 2 datacenter licenses.
|
|
#
?
Feb 12, 2018 18:25
|
|
|
stevewm posted:Maybe someone here can check my results... Depending on your budget you may be better off with a Windows Server datacenter license for each core but in any case if you want to do live migration you need Software Assurance too
|
|
#
?
Feb 12, 2018 18:26
|
|
|
I'm not sure if it still works this way still but with some low-density configurations it was better to license with enterprise which entitled you to 4 guest VMs per license.
|
|
#
?
Feb 12, 2018 19:09
|
|
|
BangersInMyKnickers posted:I'm not sure if it still works this way still but with some low-density configurations it was better to license with enterprise which entitled you to 4 guest VMs per license. If we look at Microsoft official page (https://www.microsoft.com/en-us/cloud-platform/windows-server-pricing) the standard license now covers only two vm, so up to three sets of standard licenses would be required for each host(Unless live migration is disabled). Depending on his vm layout it may be cheaper to just license win srv datacenter.
|
|
#
?
Feb 12, 2018 19:19
|
|
|
SlowBloke posted:if you want to do live migration you need Software Assurance too From what I have been told, this is only true if each host is not fully licensed for the max amount of VMs you will ever run. When using Standard, I have been told that if you license each host for the total amount of VMs you will ever run (in this case 6) then you can move VMs around as much as needed. This is also what I have managed to understand with reading the documentation from MS.. Since host #1 would be licensed for 6 VMs, and host #2 for 6 VMs, but we will never run more than 6 Windows VMs total, then we would be OK. The SA benefit would only apply if each host was only licensed for 4 VMs each, and I needed to move all the VMs to one host. Without SA you can do this once every 90 days, unless it is a hardware failure. I was going with Standard because I cannot see any situation where we will ever run more then 6 Windows VMs, given what this cluster will be used for. Standard should be cheaper from what I can see. The break-even point for Datacenter is commonly stated as 13 VMs. Since everyone I ask seems to have different interpretations on this (including some VARs!) , I figured I would ask. stevewm fucked around with this message at 19:41 on Feb 12, 2018 |
|
#
?
Feb 12, 2018 19:29
|
|
|
stevewm posted:From what I have been told, this is only true if each host is not fully licensed for the max amount of VMs you will ever run. Hmm our VARs and Microsoft itself stated that you can migrate the vm every 90 days or have SA, making it kinda kludgy unless you lock the VM in place but as you said Microsoft position on the topic is very fluid. The latest virt licensing guide can be found here http://download.microsoft.com/download/3/D/4/3D42BDC2-6725-4B29-B75A-A5B04179958B/WindowsServer2016VirtualTech_VLBrief.pdf if you want to check the latest golden standard. Dunno about your licensing status(business/gov/edu/etc.) but our Datacenter licenses costs about as two sets of standard licenses, a three set of standard might be more expensive. Check at least a couple of quotes before committing to standard.
|
|
#
?
Feb 12, 2018 19:51
|
|
|
SlowBloke posted:Hmm our VARs and Microsoft itself stated that you can migrate the vm every 90 days or have SA, making it kinda kludgy unless you lock the VM in place but as you said Microsoft position on the topic is very fluid. The latest virt licensing guide can be found here http://download.microsoft.com/download/3/D/4/3D42BDC2-6725-4B29-B75A-A5B04179958B/WindowsServer2016VirtualTech_VLBrief.pdf if you want to check the latest golden standard. I've seen that document... And I think a key line is this: Microsoft posted:For Windows Server software, except in a few cases, licenses may only be reassigned to new hardware after 90 days. Which is what I am basing my interpretation on.
|
|
#
?
Feb 12, 2018 20:04
|
|
|
stevewm posted:I've seen that document... And I think a key line is this: I double checked the document(I won't deny going by memory rather than reading it before posting  ), it looks like SA is no longer required for live migration. If the vm number is not going to increase, your license BOM looks fine. I checked one of our Microsoft resellers and the license breakpoint is currently six(after six sets of standard licenses it's cheaper to run Datacenter) if you wanted a exact number for future reference.
|
|
#
?
Feb 12, 2018 20:22
|
|
|
SlowBloke posted:I double checked the document(I won't deny going by memory rather than reading it before posting Thank you for the sanity check. I had 2 VARs telling me one thing (what I had interpreted), and another telling me something different. Grr.... Why do they have to make it so drat confusing. The licensing for this project costs more than the hardware it will be running on. Needing 145 RDS User CALs really hurts...
|
|
#
?
Feb 12, 2018 20:28
|
|
|
Part of my desire to just run everything in  is to give up dealing with the world of endless dogshit that is MS licensing
|
|
#
?
Feb 12, 2018 21:37
|
|
|
stevewm posted:Grr.... Why do they have to make it so drat confusing. Because of the VARs. Insane licensing is kind of like "professional install only" A/V equipment. It's not actually about helping the customer, it's about keeping a role for another needless middleman.
|
|
#
?
Feb 12, 2018 22:31
|
|
|
wolrah posted:Because of the VARs. Insane licensing is kind of like "professional install only" A/V equipment. It's not actually about helping the customer, it's about keeping a role for another needless middleman. Nah, Microsoft doesn’t give a poo poo about VARs, the licensing is the way it is because they determined that was the best way to maximize profit and push people towards Azure. I don’t think any VAR actually likes Microsoft licensing being arcane bullshit because it generates a bunch of calls and meetings that aren’t billable. The margins are the same whether the licensing model is sane or not.
|
|
#
?
Feb 12, 2018 22:38
|
|
|
YOLOsubmarine posted:Nah, Microsoft doesn’t give a poo poo about VARs, the licensing is the way it is because they determined that was the best way to maximize profit and push people towards Azure.
|
|
#
?
Feb 13, 2018 05:08
|
|
|
I have a Fedora/Linux host running on a ThinkPad with a Windows 10 guest for Excel/Office. I’ve been using Excel more to the point where I just need to keep the Windows VM spun up all the time. What options/settings can I change to make this VM experience as smooth as possible? This is running on a quad core Haswell i7. I’ve given it 2 cores so that the host has two dedicated cores still. I’ve also given it 8GB of RAM and the 3D video RAM is maxed out and 2D/3D acceleration are turned on. The most graphics intensive app I’d run is PowerPoint so do I really need those enabled? I’m also not adverse to switching from VirtualBox to something else if the performance gain is trivially easy. Or also just switching from Windows 10 to 8.1 if that works out better too.
|
|
#
?
Feb 13, 2018 06:58
|
|
|
Boris Galerkin posted:I have a Fedora/Linux host running on a ThinkPad with a Windows 10 guest for Excel/Office. I’ve been using Excel more to the point where I just need to keep the Windows VM spun up all the time. If it's a Windows 8 or 10 host i would suggest moving to the integrated hyper-v instance which is quite faster than virtualbox for Windows guests.
|
|
#
?
Feb 13, 2018 07:26
|
|
|
|
| # ? May 14, 2024 00:26 |
|
|
SlowBloke posted:If it's a Windows 8 or 10 host i would suggest moving to the integrated hyper-v instance which is quite faster than virtualbox for Windows guests. How do I do that? Wikipedia says Hyper-V is a Windows thing and I absolutely need a Linux host. e: After reading about it I’m not sure if I absolutely need a Linux host. I do a lot of numerical/computational stuff, so it’s pretty much been Linux all the way down because I don’t want to recompile libraries/tools for different architectures, and also because all the computers I use are also Linux so it just makes everything easier. But if I can run a Windows host with headless Fedora/Linux with zero performance hits while letting me develop MPI applications when I run them in the Linux VM then I guess that might work. But this also seems like a lot more work than asking if there were obvious settings that I could flip on/off to gain free performance for Excel, which I don’t even use for number crunching. I just want a lightweight Excel app, and LibreOffice is poo poo and Google Sheets and Office Excel Online are too slow. Boris Galerkin fucked around with this message at 08:53 on Feb 13, 2018 |
|
#
?
Feb 13, 2018 08:43
|
|