@Jimmy9008
Taking into account the cluster specification you mentioned, StarWind is not the bottleneck as the network configuration might be. For the similar setup of StarWind HyperConverged Appliances, we install, at least, 25 GbE network adapters and switches, however, 40 GbE would be preferable. I believe you should benchmark the network performance, whether 10 GbE network is the bottleneck or not for your environment.

As it was emphasized, use Live Optics to get current performance utilization picture by your production. Additionally, you might use diskspd to benchmark storage performance and network utilization.

Please tag or dm me should you have any questions.

Have a nice day!

The Intel's S4610 that @Jimmy9008 referred to in his original post, is sold by Dell as well. We have a couple of Dell servers on order with those drives (Dell branded).

I wouldn't use consumer SSDs in a server at all. They don't have power protection and not as good wear leveling. And the price difference between the better consumer drives and the read intensive enterprise drives is not big.

Anyway, the only problem I see with buying the Intel drives from somewhere else and not Dell, is that they will be under Intel warranty and support and not Dell. So if you need 4h mission critical support on those drives, you better have a spare on the shelf instead.

Dell has to stock them for warranty and Dell has to make money of them and the price have to be high enough so that Dell can give their large customers a big discount and still have some profit left. That's why the Dell branded drives are much more expensive.

Why not buy two S4610 drives in smaller capacity like 240GB or 480GB and put them in a server and see how it works?
Looks like two 480 GB drives will be around £300.

A lower cost alternative is the S4510 drives. They can't handle as many writes as the S4610 but will still be overkill for most workloads.

@Jimmy9008 said in CSV... what happens at a lower level?:

That makes sense, would a drop from 1GB/s to 100MB/s be expected? Seems huge...

Depending on the system, yeah, especially with certain kinds of operations.

@Pete-S said in ASA 5516-X Intermittent Downtime:

On the contrary, it tells us a lot about the business needs. Since the business decided to call him in the middle of the night, someone decided the firewall was important enough for them to do that, instead of waiting until the morning.
Maybe nobody brought up the HA option when the firewall was put in place or the need wasn't there at the time.

That someone was willing to call to report an issue truly tells us nothing. I get called off hours about truly worthless things all the time. Just because one person thinks that they should call and knows how to call doesn't mean that someone has assessed value. But that someone didn't buy HA tells us that at least at some point, someone decided HA wasn't worth it. That might have changed, and maybe they were wrong even at the time, but that decision of HA or no HA was made and a design built around that.

The business isn't who necessarily called him, that denotes a key decision maker representing the company. All we really know is that someone working at a site decided to notify him. Fall all we know that was an intern who just happened to find his phone number. Or it was the CEO, we just don't know. That someone decided to call because they noticed something has to be taken with a grain of salt as we don't know who they were, or what they expected. Maybe they thought support was 24x7 and that it was their job to report things by phone whenever they happen. That's not uncommon and wouldn't give us any insight into the business' evaluation of the need.

Spanish or English would be fine. But, needs to be local to Barcelona. The only reason we are looking is for the physical side we cannot do from London or Calgary.

@Jimmy9008 said in VMQ issues/Veeam/Windows Server 2019...:

@dafyre said in VMQ issues/Veeam/Windows Server 2019...:

@Jimmy9008 said in VMQ issues/Veeam/Windows Server 2019...:

@dafyre said in VMQ issues/Veeam/Windows Server 2019...:

@Jimmy9008 said in VMQ issues/Veeam/Windows Server 2019...:

@dafyre said in VMQ issues/Veeam/Windows Server 2019...:

@Jimmy9008 said in VMQ issues/Veeam/Windows Server 2019...:

Hi folks,

I am having VMQ issues on my 3 x Windows Server 2019 host. I am using a team of Intel X550 NIC (3 x 10 GbE interfaces to 3 x Stacked switch). The Team has been configured using Switch Independent mode.

Now, VMs are on the LAN perfectly and communicate. Rock solid.

Now to the issue. When I use Veeam with Application Aware turned on, the hosts get stuck creating checkpoints for VMs at 9%. The only option is to kill the host, restart and turn the VMs back on. It happens every time. (Strangely, checkpoints run fine native).

Veeam have looked at the logs and have said its an issue they have seen before, and to turn off VMQ initially on NICs and VMs as that often solves the issue. No worries. I do that, and the backups then run fine - perfect!

However, VMs over time then drop off of the network. I can connect to them in Hyper-V, but nothing I do will bring them back on the network. Initially they are on the network, just at some point in time many drop off, whilst others stay on.

The only resolution is to turn VMQ back on and reboot. I cant really keep testing this either as it causes much downtime! Not good. Of course, when VMQ is back on... Application Aware backups then fail and kill the host like I said at the start 😕

Any idea why some VMs drop off of the network with VMQ disabled on the VM and NIC? Host, fully patched. NIC, latest firmware. I thought you didnt have to use VMQ...

Best,
Jim

Is the the same VMs that drop off the network every time?

What OS are the VMs that get dropped off the network?

What kind of switches are you running?

Uncertain. I've noticed a few are the same, but I've had no time to really look in to it. Too many users. I'm trying to get ideas that could be the issue to do proper testing with downtime around Dec 20th when a lot are on holiday.

I think 2012 R2. Possibly 2016 too. Can't recall any 2019.

Dell N4064 Stack.

Try the ArpRetryCount trick. You'll have to reboot each VM or host after you make the change.

Ok, I can try that. So:

disable VMQ physical disable VMQ on all VM NIC change registry reboot

Sound about right?

change the registry on any machine that is down.

But yeah, that sounds about right.

Do I need to change the reg on the host too?

I would go ahead and do it, yeah.

@JaredBusch said in Veeam Active Full or Synthetic Full?:

Because I sync Veeam’s backup target (a Synology NAS) to B2, I do not use the synthetic full that daily recreates the full backup. That would cause the entire vdk to need to be uploaded to B2 every day.

In my case I set up active full to run on Fridays. A different server each Friday. Stacking various serves on the same Friday to balance the load over the month.

Then daily incrementals.

This is also exactly why I don't do this.

It looks like you'd probably want ZRS

https://docs.microsoft.com/en-us/azure/storage/common/storage-redundancy

@scottalanmiller ehhh, this isn’t a university It’s a trade school (oddly one that certified people in CCNA, helicopter repair etc). It’s a decent trade school but I suspect like a lot of universities an schools in rural areas they count of low job competition for non-remote entry level work. For what it’s worth a large 4 year university on the Brazos was paying only 40K for The head of ResNet department.

Universities priorities are strange. From a compensation plan free student tuition, and free masters often mean they can treat a lot of salary positions line work study jobs from a comp basis.

For what it’s worth I’ll likely be (jetlagged) but back in Waco for the OU game (or any games after it). I’ve made it to only one game this season sadly.

I have friends who still live in Waco but they all work remote. There’s a decent work from home Wordpress community there. The job market isn’t great for IT infrastructure unless you have clearance but even then [Redacted] IT is so frustrating wild horses couldn’t drag me back into one of their offices.

Just move to Houston or Dallas or another market until you can skill up enough to work remote.

@Dashrender said in Cisco ASA:

@Jimmy9008 said in Cisco ASA:

A and B can also RDP/ping devices sitting on C.

If this is true, it's just a matter of rules/route allowing C back to A/B or a route specifically for C -> A/B.

172.16.0.0 vlan… switch IP = 172.16.0.1, ASA = N/A, gateway on the vlan is 172.16.0.1 (the switch)

this is legacy. What appears to happen is that the switch has 0.0.0.0 set to 192.168.50.10 (the ASA) on a vlan2. So, traffic from 172.16.0.0 hits the switch IP at 172.16.0.1, then hope out 0.0.0.0
^ I think its this that's causing the issue.

This should be fine, this is what allows the C network to get to the internet

so, when on the 172.16.0.0 network, the request goes to the switch's IP (172.16.0.1) which forwards it to 192.168.50.10 (the ASA), The ASA then doesn't have a rule allowing traffic from 172.16.0.0 to talk to 10.x, so it just dumps the traffic.

At least that's what it looks like to me at this time.

“C” network really?

@galezer said in EMC VXRail:

if you count snapshots usage and price vs vmware and you are familiar with KVM its a win/win ...

Sure, but it's a false comparison. The main competitors are Scale, Starwind, Simplivity, etc. They crush Nutanix in performance, ease of use, and price. Comparing Nutanix against a known bad alternative (3-2-1 with VMware) makes it look good, but that's like saying paying $100K for a Chevy Spark is a good deal because a Ferrari costs $300K. But that's not a good comparison. A good comparison is a $12,000 Nissan Versa which costs a reasonable amount and is still better than the Spark.

It's true that the Nutanix is likely an improvement over an inverted pyramid + VMware, both things you'd normally rule out as just bad design. Using a false comparison makes it look plausible. But when you compare it to industry baselines of "standard good" it stands out as expensive and under performing (and just a bad actor as a vendor.)

You need to put a route on the ASA so it knows how to get to VLAN 6

something like

route add 10.12.10.0/24 10.12.0.2

@Dashrender said in License Hyper-V Virtual Machines...:

@Obsolesce said in License Hyper-V Virtual Machines...:

@Jimmy9008 said in License Hyper-V Virtual Machines...:

@Obsolesce said in License Hyper-V Virtual Machines...:

@Jimmy9008 said in License Hyper-V Virtual Machines...:

Hi folks,

I have Windows Server Datacenter 2019 licenses. If I install Hyper-V server, rather than Windows Server with the role, can I setup AVMA and have the host auto license the VMs?

Or, do I need to login to each VM and add the Datacenter key?

Best,
Jim

Avma does not work on Hyper-V Server.

You'll need to install Hyper-V role on Windows Server DC if that is a must have among any other DC only features.

So, if I have the Hyper-V Server installed only - how do I use my Datacenter license on the VMs? I guess I login to each and enter the key manually?

Correct, or you can do this instead:

https://docs.microsoft.com/en-us/windows-server/get-started/kmsclientkeys

This is the recommended solution.

You have a DC license, so you can spin up a VM that does nothing but act as a KMS for your other server VMs.

Right with DC, you have no issues with quantity of licenses.

@Jimmy9008 said in IBM Storewize V3700 - Supported Drives:

@scottalanmiller said in IBM Storewize V3700 - Supported Drives:

D) drives:

So, any SAS SSD under 1.6 TB should work with this? Perfect.

That ALWAYS refers to Lenovo's own drives. It never implies third party drives. This is true for all server vendors.

@Jimmy9008 said in Wireless headset for Skype for Business...:

When away from the desk, the headset will ring if somebody calls in, but you are not able to answer the call from the headset. So, you have to run back to the desk and answer the call in Skype for Business soft phone, which then connects the call and works over the wireless headset.

This isn't a Skype for Business issue - this is just a lack of pickup button on the headset issue, or is there a button that works with other software, just not S4B?

As mentioned - real professional gear like Sennheiser and Plantronics work well in this space.

@larsen161 Yes, I get that. 🙂

Yes, if you are setting physical ports to a VLAN, then they are acting like a physically different switch on those ports. So attaching another physical switch to one of those ports would make it a switch on that VLAN.

Correct. You have to have routing to make this work.

Every router OS implements it differently.

Hell the EdgeRouter does it multiple ways if you need to.

A basic IPSEC tunnel where you simply specify the local and remote subnet, one pair per tunnel (each IPSEC connection can have more than one tunnel pair). Or a more advanced vti interface where you put in specific routes.

@PhlipElder said in Raid10, must use or another Raid limits?:

Modern RAID controllers have the horsepower and cache RAM that is flash backed to overcome any real parity performance costs.

Operationally it would be just fine. It's only recovery time that would be of concern.

I got some drives from xByte that did not work in the DELL server. They WORKED, but kept freezing, or rebooting. xByte, of course, were awesome and swapped them out with DELL branded ones.

For me, I’ll never put on a lifejacket again. Oh wait, wrong line. I'll never not buy DELL branded stuff for my DELL server.