Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
Mr Shiny Pants
Nov 12, 2012

adorai posted:

1) Get 2x HA pairs of Oracle Sun 7320 storage.
2) Get 4x (2x for each location) Cisco 3750x gigabit switches
3) License everything you need

For #1, I went all out with ours and got roughly 15TB (usable) on each with tons of cache for $100k for both pairs. You could get less storage and less cache and probably end up somewhere around $80k
For #2, I think you can probably do all four switches for well under $10k. They stack and allow you to do cross switch etherchannel links, giving you good enough speed and redundancy for your organizations needs.
For #3, I can't exactly comment.

I think you could do the entire project for under $100k. You would be using only gigabit Ethernet rather than 10gig Ethernet or FC, but you can replicate between the two.

How are these working out for you? Is the performance good? Have you tested HA on them? I am curious because I am a huge ZFS fan and some real world experience with Oracle/ Sun gear would be nice.

Adbot
ADBOT LOVES YOU

Mr Shiny Pants
Nov 12, 2012
That's great to hear, does Oracle have a roadmap for the systems? I don't want to buy/advise something with no upgrade path....

Is this your primary storage? If not why not?

Mr Shiny Pants
Nov 12, 2012

evil_bunnY posted:

The problem with oracle ZFS is that 80% of their core engineers left after they closed Solaris.

Not to defend Oracle or something but isn't this the problem almost anywhere? EMC has lost a couple of their directors who started another company and build the XIV. NetApp is laying off 800 people and other tech companier are doing the same.

Mr Shiny Pants
Nov 12, 2012

Misogynist posted:

The bigger problem is that the divergent ZFS codebases mean that many more independent vendors are using Illumos ZFS than Oracle ZFS. Oracle now has vendor lock-in on their formerly open-source filesystem, while the competition doesn't.

This typically doesn't factor into business decisions, but it's important, I think. Especially so now, since open-source ZFS uses feature flags instead of hard/meaningless version numbers.

True, Version 28 is going to be the best version I guess.

Mr Shiny Pants
Nov 12, 2012
I've been looking into OpenStack a lot recently and from what I've seen the technology is pretty good but, like others have said, pretty rough around the edges.

It's built for VMs that are not long lived. For many businesses this will probably not be a good fit.

I've also been looking at Cloudstack and that looks much better. Combined with Ceph it seems like you can built a pretty awesome cluster on relatively simple hardware.

Mr Shiny Pants
Nov 12, 2012
L2ARC is the caching mechanism that ZFS uses.

You create a pool of disks and to that pool you add a cache disk.

ESXi only supports one physical processor. KVM here you come ;)

I run Ubuntu 13.04 with KVM and ZFS. Works pretty awesome.

Mr Shiny Pants
Nov 12, 2012

jre posted:

The free version of esxi has no limits on physical cpus, The essentials versions have a two socket limit.

http://www.vmware.com/products/datacenter-virtualization/vsphere-hypervisor/requirements.html

Well scratch my remark. Thought i read it somewhere.

Mr Shiny Pants
Nov 12, 2012
I have an 128GB SSD in my main machine and I have a ISCSI drive hooked up to my ZFS box for Origin and Steam. 83GB of games is a bit much on an 128GB SSD.

No messing about with programs that don't understand UNC networkdrives. It behaves like a disk and works perfectly.

When Battlefield loads a map it sometimes goes up to 200Mbps.

This is on Ubuntu running LIO.

Mr Shiny Pants
Nov 12, 2012

dont change my name posted:

What's a good libvirt front end for KVM? Something nice and easy with lots of gui and charts and things.

Ovirt?

Mr Shiny Pants
Nov 12, 2012

Kachunkachunk posted:

You're my favourite grump in this thread.

The webclient looks awful next to the clean vsphere client. It looks like SCVMM, which feels cluttered.

Mr Shiny Pants
Nov 12, 2012

Dilbert As gently caress posted:

They work basically the same, only IIRC hyper-V has some of the built in plugins install in it's tools for VSS on exchange and SQL servers to quiesce the filesystem and DBS.

Does anyone have any real experience with this? We have some challenges with VMware that, due to the way its snapshotting works, we can't make consistent snapshot as many times as we'd like because of the VM stuns when it starts replaying the redo file into the VM after a snapshot.

I read that HyperV works differently and does not have this problem but I haven't found anything conclusive.

Mr Shiny Pants
Nov 12, 2012
Install Windows 8 and use HyperV? Do you have a SSD in it?

I think Hyper V performs much better than VB. An SSD is always a good choice.

Mr Shiny Pants
Nov 12, 2012

Knighty21 posted:

Is the vmware box located on the same PC you're accessing it on?
Intuitively it seems like adding an RDP layer would just create more overhead.

RDP is very efficient. The console is like emulating the vid card.

Mr Shiny Pants
Nov 12, 2012
I don't exactly know where to put this but here goes:

Does anyone know if Linux bridging works under Hyper-V? I've been trying to setup bridging for LXC and I am having a hell of a time getting the containers to work.

I've tried the following Linux sysctls:
net.bridge.bridge-nf-call-arptables = 0
net.bridge.bridge-nf-call-ip6tables = 0
net.bridge.bridge-nf-call-iptables = 0

To disable any filtering but I still can't get DHCP to work or ping another machine (but the Linux host ) from a container.

I've created a br0 with brctl and everything looks right, but it does not work.

This is making me doubt the networking part of Hyper-V.

Any clues?

Mr Shiny Pants
Nov 12, 2012

Thanks Ants posted:

Is this any good?

http://www.cloudbase.it/hyper-v-promiscuous-mode/

I had to do something similar in vSphere to get a VPN concentrator to work properly.

No dice, I hoped that would be it but it does not seem to work.

It's also pretty horrible to pinpoint where it goes bad.

If anyone else has an idea, please let me know.

Mr Shiny Pants
Nov 12, 2012
Well my problem of Linux bridging not working in Hyper-V turned out be a simple checkbox. Turn on "Enable MAC address spoofing" in the advanced features of the VM nic and BAM! it works.


Now to get that day back I spent troubleshooting this....... :(

Mr Shiny Pants
Nov 12, 2012

Misogynist posted:

All you lollygagging young'uns complaining about Microsoft licensing need to try pricing a datacenter-wide IBM product rollout by Processor Value Units.

Or Oracle. "Is a core 75% of a normal processor or is that hyperthreading". pfffff

Mr Shiny Pants
Nov 12, 2012

evol262 posted:

Imagine IBM pricing every POWER system at 1 processor unit so you get a deal on Tivoli, then turning the screws as you slowly migrate to x86, with every additional SMP core being counted, then every additional SMT core, so it's 2012 and you can't even buy a server with less than 8 "cores", so your licensing costs have gone from $1M/yr to $15M/yr,

See boneheaded moves. And they wonder why people are not buying IBM anymore. We had Tivoli backup, my god the per TB licensing was insane.

Mr Shiny Pants
Nov 12, 2012
Are there any gotchas for Vsphere 5.5 update 2 that I need to be aware off? We are updating our cluster at the end of the week because of some VSS errors and some other things relating to the VMCI driver.

Mr Shiny Pants
Nov 12, 2012

evol262 posted:

It's because they're developers who have never worked as admins/engineers dealing with it. And we don't consider VMware competition. Product management certainly does, but what VMware is or isn't doing doesn't really matter to us when we're working, individually.

"Gluster and oVirt hyperconvergence" certainly is a analogous to vSAN. Whether the guys doing the implementation know that is another question. How VMware does it doesn't matter to us at all, in any way, since we're using totally different tools and we couldn't re-use their architecture even if we wanted to, so why learn it?

Because the terminology is useful, if people come from a MS shop they will ask about Active Directory like features. It is easier if your people know that they are talking about centralized management of users and some SSO stuff instead of "No, we don't do that."

It makes it easier for customers if you can compare stuff by its name. IMHO.

Mr Shiny Pants
Nov 12, 2012
Double post?

Mr Shiny Pants
Nov 12, 2012

EuphrosyneD posted:

Newbie question about automatic virtual machine activation in Hyper-V ahead.

Can I build the machines I want to activate on Windows 8.1's Hyper-V ahead of time, then migrate them to another Hyper-V host based on WS2012R2 Datacenter, and expect the activation to work?

Or do the guests have to be built on the host they'll activate on?

The hardware should be the same, Hyper-V in 8.1 is the same as in 2012R2 so it should work.

One of the use cases MS talks about is developers migrating their VMs to their local 8.1 workstations from Hyper-V clusters, so theoretically you should be fine. It would be dumb that in scenario like this you would need to reactivate the VM. Could be that you would need a KMS server instead of MAK keys though.

But, as with everything, I would test it first. :)

Mr Shiny Pants
Nov 12, 2012

KS posted:

That is loving stupid. Question everything else you're "learning" from him. It's actually critically important in an environment where L2 is spanned across sites, because you want to be very careful and deliberate about traffic going across your WAN.


Sorry, but so is this. DCs are highly available -- just stand up another one in the other site. Use a DAG for Exchange. Active/active clusters are pretty complex undertakings: you need spanned L2 networks, and you need a redundant highly available WAN since you're running storage traffic over it. Candidly, you don't sound qualified to design this stuff if you're questioning the need to subnet.

A single site cluster gives you this as long as you have N+1 capacity. It's kinda the whole point.

If the WAN is just a leg in the network, an extension of the normal network, why would you need to worry about traffic going over it? You want it as transparent as possible. Granted you would put servers that need services from each other close together.

You can have affinity with servers located close to where your users are so as to not make the WAN a bottleneck.

We have a 100 mbit linkup to another DC that also runs the same IP space, so in the event of a failover we can just power-on servers without needing to worry about reassigning IP addresses or DNS entries.

If the connection is reliable, get two at least from different providers, I don't see the need to split-up your network and make things harder for yourself in the event of a fail-over.

Mr Shiny Pants fucked around with this message at 18:22 on Apr 3, 2015

Mr Shiny Pants
Nov 12, 2012

Erwin posted:

You say this as if IT companies aren't known for charging for software features.

Cisco Nexus VM anyone?

Mr Shiny Pants
Nov 12, 2012

SlayVus posted:

Using the dells as clients, not to run virtual machines.


According to the tech specs, the LAN supports PXE.

Since these are actual computers, would I still go about assigning them resources from the host or could they use their own CPU? There by reducing the hardware requirement of the host machine.

No, the point is you use server resources. The Dell machines are just dumb clients performing the least processing necessary.

Mr Shiny Pants
Nov 12, 2012

Internet Explorer posted:

Yes, Hyper-V on Windows 8 allows you to create VMs.

It'll even do Gen2 VMs which work with most Linux and newer Windows versions. It is pretty nice and really, really fast..

Mr Shiny Pants
Nov 12, 2012

madsushi posted:

So the web client guys have moved on to containers, got it.

Nice :)

loving webclient :(

Vulture Culture posted:

Here's everything you need to know in a few paragraphs.

snip..

This is great, thanks.

Mr Shiny Pants fucked around with this message at 10:00 on Jun 12, 2015

Mr Shiny Pants
Nov 12, 2012

DevNull posted:

Hopefully the success of the embedded host client (I hate the name) will convince the higher ups that we need to focus on html5 and dump flash fast. I've been watching this trainwreck for something like 5 years now.

So if you want to see that succeed, leave a comment on the Fling page saying how you hate the web UI, and how much you love the embedded client.

Can we still also love the C# client? Please....

Mr Shiny Pants
Nov 12, 2012

DevNull posted:

It has no future. It was kept on life support only because of the problems with the web client. The embedded client is more likely to get good support than the C# client.

Weird how a company that says that it listens to it's customers will discontinue a feature that is well loved by the same customers.

Mr Shiny Pants
Nov 12, 2012
Wow, this might be something I will try. Windows 10 is not for me, but I do have some games I want to play. If I could run a VM with decent framerates and use a Steam Link to display it on my TV.....

Mr Shiny Pants
Nov 12, 2012
My TS440 has an PCIe 16x lane slot and I have a spare 660GTX. Might try something in a couple of days.

Mr Shiny Pants
Nov 12, 2012
I don't think that is possible, maybe with some scripts. But if it takes a reboot then you can just as well dual boot. The nice thing about this is that I can make my main machine a Linux or OSX box and still play my PC games without dual booting.

Luckily my TS440 has onboard video so I can just use that for the host OS and run the Windows VM with a dedicated GPU attached. It would be awesome if you could partition your vid card, didn't some Nvidia cards allow this? One third for the main OS and maybe two thirds for the VMs.

Mr Shiny Pants
Nov 12, 2012
So if I flash my 660GTX to UEFI firmware, use the OVMF bios in my VM and have my server boot in UEFI mode this should work? Cool I can try that.

The steamlink is just a way to disable having to have a physical keyboard and monitor attached.

Mr Shiny Pants
Nov 12, 2012

Martytoof posted:

It's hard to say. I mean in a perfect world you're going to want the fastest everything. Anecdotally speaking I don't really see a huge performance boost between my Dell 1950-IIIs with ECC-DDR2-whatever and my new R620 with ECC-DDR3-whatever, at least in the applications I run on my homelab, so I'm sort of going with the whole "throw your money at other parts of the system" viewpoint. I think the highest memory-intensive application I run is Splunk, then Hercules. If either has been impacted by faster memory then I doubt I'd notice.

Yeah, memory could be a few percent and is really not that noticeable. Making the choice between SSDs or spinning rust however..........

Mr Shiny Pants
Nov 12, 2012

DevNull posted:

It was definitely a holy war getting it to this point. There was a lot of political resistance along the way.

The work for it started 3 years ago with a little hackathon the company did. One of the groups did a javascript client that allowed you do do basic functions on an ESX box such as browse the datastores and manipulate VMs for basic things. Another group of 3 engineers ported the webMKS to ESX. That gave you a console to the VM. The two groups merged their projects and tried to get the company to launch it as a real project. It didn't get picked up though because the webUI group was still pushing the Adobe Flex stuff. Their solution to connecting with vSphere was install flash on a windows host, and installing some package that you could connect to ESX. It was bad, I had trouble getting it setup. In the mean time, one of the vmkernel engineers saw the hacked together javascript/html project. He convinced the vmkernel director to staff it up directly. They hired another engineer with UI experience, and together with the vmkernel engineer, the produced the Fling. Obviously the Fling did well, so it was changed to a supported product. VMware sure as hell better give the vmkernel engineer a big promotion and pay raise. He absolutely nailed it with this project.

I bet they fired him, kicking up a storm like that. I hope not, but I would not be surprised.

Mr Shiny Pants
Nov 12, 2012

DevNull posted:

No, he is good. The director that was backing him is now actually a VP, so he has even better protection.

Cool :)

Mr Shiny Pants
Nov 12, 2012
^^ this. This whole devops poo poo is getting out of hand.

Mr Shiny Pants
Nov 12, 2012

HPL posted:

I just figured out there's a Windows QXL driver on the virtio Windows ISO. What a difference. Now video performance is smoother and mouse response is way, way better.

I just use RDP.........

Mr Shiny Pants
Nov 12, 2012

NippleFloss posted:

Agreed, VEEAM is the best option here. Volume based replication will require something like SRM to automate the adding to inventory and re-ip work. Veeam replication was built to do this.

Veeam is great, testing your DR/replicated VMs in an isolated network to see of they really do come up is really really slick.

Adbot
ADBOT LOVES YOU

Mr Shiny Pants
Nov 12, 2012
I can't imagine people willingly buying new Oracle stuff when given the chance. I'd rather not use software than have to deal with Oracle.

"Yes I would gladly pay a lot of money to a vendor and get abused everytime they think of some new licensing trickery."

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply