Hyper-V 2 node cluster replacement

[lmgtfy]

@CAWJames It might be worth looking into DataCore https://datacore.co.uk/ You don't have to spend the earth getting a two node Hyper-V cluster and we looked into it with a company called M-Tech Systems. https://www.mtechsystems.co.uk/

 

Quite a few schools look like they use DataCore according to their page. M-Tech said they could even supply it using refurbished Dell servers to reduce the price.

 

Cheers

[Paid_Peanuts]

We have a DR site at another centre where all our DPM backups are collected by a secondary DPM system, where we can spin up all the VM's in the case of the main site going dark, then we have round robin on our DNS for all our external IP's. That's the 1 working day RTO. Restore from backup and reprogram watchguard to point all the external IP's to the "new" servers, as at the moment, they all point to the main site, so round robin doesn't give us nothing found 5 times out of 10

 

CAWJames - can you give us an update on your project? Have you made a decision either way on your SAN replacement?

We are in the same position, our 3 hosts (VMware ) and EqualLogic (PS6110x, PS4110x and PS4110e) are due for replacement in March 18.

I am just considering our options.

 

Im open to other vendors and type of solution, I.e. hyperconverged, vsan/storage spaces etc.

Edited November 24, 2017 by Paid_Peanuts

[CAWJames]

CAWJames - can you give us an update on your project? Have you made a decision either way on your SAN replacement?

We are in the same position, our 3 hosts (VMware ) and EqualLogic (PS6110x, PS4110x and PS4110e) are due for replacement in March 18.

I am just considering our options.

 

Im open to other vendors and type of solution, I.e. hyperconverged, vsan/storage spaces etc.

 

Sure, current position is:

 

We have looked at the solutions that would fit our use case, mainly:

1. Shared Nothing Live Migration with 2 hosts each with it's storage - discounted due to the cost of doubling down on storage and speed of the migration
   
2. Standalone hosts - Discounted due to inflexibility, lack of resilience, etc, etc
   
3. Hyper-converged - While it would be nice to just add computer or storage nodes as when needed, for our use, the initial cost would be extreme overkill, and did I mention the cost?
   
4. Good ol' SAN and hosts - Basically refresh like for like. This is the current step we are sitting on.
   
5. 
   

I did my usual talk to vendors, ask for oak trees, offer peanuts, and waited for the quiet sniggering at the end of the line. Wants are 2 beefy hosts with 256GB and dual xeons and a SAN that can handle 20TB after RAID6, 10Gbe or other fast links, and we want to spend under £20k...

 

We are still getting quotes in, but at least one of our vendors came back with a "no problem we can do that", I nearly fell off my chair.

 

So that best quote, to paraphrase. has 2 servers with E5-2630v4 Xeons, 256GB DDR4 RAM, 2 x 300GB 10K 12Gbps SAS drives, Dual PSU, 10Gbe. And then a SAN with 16 x 1.8TB 10K SAS and 4 x 800GB SSD drives, an SSD tiering licence and 10Gbe connectivity for around our target figure, but I expect to beat the crap out of the price to get it to our preferred level. All hardware is Lenovo.

 

That's not much more than what we paid for our Equalogic PS6110x 5 years ago on its own!

 

We haven't made any hard decisions yet, but that might be a goer for us, we are still waiting for some more quotes and pricing to come in though.

 

Hope that helps some?

 

James

[Guest Guest]

Have you looked into the Dell VRTX? It’s basically a blade enclosure with DAS built-in and space for either 4 or 6 blades. You can get some refurbished gear within you budget

[nicholab]

Also Lenovo have an offering with datacore for a hyper convergence system. You can use there software in the cloud also you could reuse your current storage with a datacore cache boxes.

[AlanD]

There seems to be a conception of a nirvana where ICT can guarantee continuity of service by use of virtual servers, multiple SANs, dual server rooms, duplicated core switches, etc, etc.

 

And somewhere in this mix is the illusion that SLA is going to get you up an running in a few hours - whereas the reality is that despite an engineer being on site within that time there is no guarantee - not even a financial one - that a fault can either be fixed or you can then recover your working systems. Some systems require a whole day just get their data back from a backup.

 

The reality is that even the biggest IT companies in the world, despite armies of highly skilled and experienced managers have had significant outages - sometimes extending to several days of disruption.

 

So sometimes I think a better plan is not to promise what is often an unrealistic guarantee of service. Some house two physical servers, SAN and backup server in same building - sometimes the same server room...with same UPS....but justified the need for two servers to give redundancy should one fail....

 

Either you need to be really serious about redundancy - and that will be expensive (and still not 100% guaranteed) - or you may be better off explaining the risks to management that you could spend another 100K but you still couldn't guarantee an outage would not take a couple of days to recover....in which case they won't be alarmed or disappointed when it goes pear shaped.

 

And even if you managed 100% physical integrity - you have a much more significant risk of malware or ransomware taking down your system leaving management asking you why you spend an extra 100K.

[PotNoodleTech]

With a 20k budget I would be favouring a VMware VSAN setup with 3 hosts. All the benefits of a SAN, but with much better redundancy, and much easier expansion.

 

The licences are not all that cheap but you get what you pay for, VSAN is way ahead of storage spaces.

 

In my humble opinion, SANs are old hat, out of date technology I'd never spec one on a new system now.

[PotNoodleTech]

There seems to be a conception of a nirvana where ICT can guarantee continuity of service by use of virtual servers, multiple SANs, dual server rooms, duplicated core switches, etc, etc.

 

And somewhere in this mix is the illusion that SLA is going to get you up an running in a few hours - whereas the reality is that despite an engineer being on site within that time there is no guarantee - not even a financial one - that a fault can either be fixed or you can then recover your working systems. Some systems require a whole day just get their data back from a backup.

 

The reality is that even the biggest IT companies in the world, despite armies of highly skilled and experienced managers have had significant outages - sometimes extending to several days of disruption.

 

So sometimes I think a better plan is not to promise what is often an unrealistic guarantee of service. Some house two physical servers, SAN and backup server in same building - sometimes the same server room...with same UPS....but justified the need for two servers to give redundancy should one fail....

 

Either you need to be really serious about redundancy - and that will be expensive (and still not 100% guaranteed) - or you may be better off explaining the risks to management that you could spend another 100K but you still couldn't guarantee an outage would not take a couple of days to recover....in which case they won't be alarmed or disappointed when it goes pear shaped.

 

And even if you managed 100% physical integrity - you have a much more significant risk of malware or ransomware taking down your system leaving management asking you why you spend an extra 100K.

 

PS this is absolutely spot on. The law of diminishing returns applies here more than anywhere, you can quickly find yourself spending 100k just to try and chase an extra 0.05% uptime across all of your "what if" scenarios.

 

At some stage you just gotta sit back and say "is this really worth the cost"? For large colleges and unis, probably, for schools, no, no I don't think it is. Not when you can achieve 99% uptime for £10k just with a pair of HyperV hosts. Chasing those extra 0.1% chunks becomes insane.

 

I find a good thing to do is actually to say to yourself - right, what has actually failed here over the last 10 years and make a list. Have you had a disk drive fail? Have you had a power supply fail? Have you had an entire raid array fail? have you had a server motherboard fail? Have you had a core switch fail? Have you had a UPS fail, etc etc, We've all had a drive fail, we've maybe had a power supply fail, what about the others? If the answer to most of those is no - then really your wasting money by trying to cater for every single what if's as history is showing you the chance of it actually happening is miniscule.

[bossman]

@CAWJames

 

We currently have a 2 host Hyper-V with 3 Synolgy RS3614xs+ NAS units, 2 of these having SSDs for caching plus 8 4TB Drives in RAID6 for 24TB storage these are clustered for failover using the Synology's own software and one unit for backups using Veeam on one of the hosts.

 

5 Year warranty with Synology plus technical support all for approx £4k each with drives, we coupled these with Dell R710s which have 2 x mirrored drives as the hosts and for the past 3 years have been more than happy, we run 20 various VMs of OS flavours.

 

We found a great company that helped me design and implement and offer their technical services at a very reasonable rate.

 

Synology have improved their technology vastly and offer even better units with the same warranty and expertise.... great value for money IMOH check them out: https://www.synology.com/en-uk/products/RS3617xs+

[CAWJames]

Thanks for all the replies, I haven't been ignoring, I am in an internet less wilderness after moving home, where there is no 3G let alone 4G and no broadband till the 5th! I will report back after a decision has been made, but I agree with chasing that last 0.1% never going to happen, I am happy with the redundancy(or lack of) that we have. Current have Dell looking to sell me a Compellent system to replace the Equallogic, but it sounds pricey already, and I haven't had pricing in yet!

[Paid_Peanuts]

Interesting that, Dell are pricing me up an EMC Compellent also as a replacement to my EqualLogic - let’s see how the price looks. Got HP and IBM/Lenovo also pricing me up alternatives.

[Norphy]

I’ve just ordered two SC3020s with about 20tb of usable storage, 4tb of which is flash. They came out price competitive with HPE and Fujitsu but their SANs were more feature rich, more expandable and had more flash.

 

Hopefully I’ll get them before chrimbo. I’ll let you know what I think once they’re in production if you like.

[Paid_Peanuts]

Hopefully I’ll get them before chrimbo. I’ll let you know what I think once they’re in production if you like.

 

Yeah any feedback would be great! I am not in a rush, I am looking to replace my existing kit during our Easter break at the end of March. I’m therefore taking my time and making sure we properly evaluate all our options.

[ColinP]

did you look at storagespaces 2016? if you want high availabilty this surely is the way to go, you will need more than 2 nodes to make it work properly but you would lose the price of the san so your budget would go further.

[CAWJames]

I got my Compellent quote and it is a LOT more than we paid for the Equallogic, even with 5 years of price changes(over £60K for Compellent and around £20k for the Equallogic including the vodka and tonic), it seems they(Dell) now have nothing in their portfolio that fits the "baby SAN" moniker that the small Equallogics filled. I've got EMC(wh coming into the mix tomorrow, but expecting that to be even higher...

 

We did look into Storagespaces, but it still looked like new tech that wasn't quite proven in production, and we opted for better the devil you know

[NB9457]

I've been looking at SANs too again.

 

I think it's going to end up being 3 beefy hosts, local storage and local VMs with an old server acting as a replica for file/core services. It's come to the point now where I don't think our SAN is necessary. For well under £20k, I'm looking at 3 hosts, minimum 128GB RAM each, 2x8 Core Xeons and 8x960GB SSDs each. Then carefully split all roles across the three and as mentioned, a replica and a VM backup system, that's a pretty good level of redundancy, more simplicity and much more speed given the SSD based storage.

 

Scenario: We lose a host, we fail back the file services that server was hosting to the replica, if it had a DC on it, there's another anyway. WSUS, SCCM, Antivirus Control, Foldr, things like that we can live without temporarily and these services being offline won't impact the end user. We have a VM backup, beefy hosts, if there's any VM it was hosting we need, we spin it up on another host. Couple of minutes downtime perhaps whilst we fail back services? Cost wise I think it's something the school can live with as we will probably save £10k by not having a SAN and touch wood, I've not had a server actually die on me yet and in the event it does happen, we have a plan.

 

I suppose, it's more of case of give the school the choice, costs, impacts and how the system works and the pros and cons of each and the potential impact in a disaster scenario. At the end of the day, it's impossible to guarantee 100% uptime.

[markwilfan]

I've been looking at SANs too again.

 

I think it's going to end up being 3 beefy hosts, local storage and local VMs with an old server acting as a replica for file/core services. It's come to the point now where I don't think our SAN is necessary. For well under £20k, I'm looking at 3 hosts, minimum 128GB RAM each, 2x8 Core Xeons and 8x960GB SSDs each. Then carefully split all roles across the three and as mentioned, a replica and a VM backup system, that's a pretty good level of redundancy, more simplicity and much more speed given the SSD based storage.

 

Scenario: We lose a host, we fail back the file services that server was hosting to the replica, if it had a DC on it, there's another anyway. WSUS, SCCM, Antivirus Control, Foldr, things like that we can live without temporarily and these services being offline won't impact the end user. We have a VM backup, beefy hosts, if there's any VM it was hosting we need, we spin it up on another host. Couple of minutes downtime perhaps whilst we fail back services? Cost wise I think it's something the school can live with as we will probably save £10k by not having a SAN and touch wood, I've not had a server actually die on me yet and in the event it does happen, we have a plan.

 

I suppose, it's more of case of give the school the choice, costs, impacts and how the system works and the pros and cons of each and the potential impact in a disaster scenario. At the end of the day, it's impossible to guarantee 100% uptime.

Exactly how we do it. Can't justify a San fast enough.

[CAWJames]

I've been looking at SANs too again.

 

I think it's going to end up being 3 beefy hosts, local storage and local VMs with an old server acting as a replica for file/core services. It's come to the point now where I don't think our SAN is necessary. For well under £20k, I'm looking at 3 hosts, minimum 128GB RAM each, 2x8 Core Xeons and 8x960GB SSDs each. Then carefully split all roles across the three and as mentioned, a replica and a VM backup system, that's a pretty good level of redundancy, more simplicity and much more speed given the SSD based storage.

 

Scenario: We lose a host, we fail back the file services that server was hosting to the replica, if it had a DC on it, there's another anyway. WSUS, SCCM, Antivirus Control, Foldr, things like that we can live without temporarily and these services being offline won't impact the end user. We have a VM backup, beefy hosts, if there's any VM it was hosting we need, we spin it up on another host. Couple of minutes downtime perhaps whilst we fail back services? Cost wise I think it's something the school can live with as we will probably save £10k by not having a SAN and touch wood, I've not had a server actually die on me yet and in the event it does happen, we have a plan.

 

I suppose, it's more of case of give the school the choice, costs, impacts and how the system works and the pros and cons of each and the potential impact in a disaster scenario. At the end of the day, it's impossible to guarantee 100% uptime.

 

Interesting, this was sort of my original plan. I've got questions What do you use for VM backup? What is your total capacity after RAID for each server with those bad boys, that is if you are using RAID? How do you deal with a failed host and spinning them up on another host? Do you restore from backup, and if so, how long does it take? I know how longs a piece of string, but ball park per GB? What kind of SSD are you using? Enterprise stamped or just generic with good RAID

 

My main reason for a SAN is really survive-ability of VM's across hosts for host maintenance and uptime.

 

This kind of system running SSD's and mulitple 10GBe connections, even share nothing live migration could be a goer, I wonder what the drain time would be in that config!

 

Currently my game plan is 2 hosts and 1 san, but 3 hosts would take the san and it's proprietary parts out of the picture, cheaper support too.

 

hmmmmm

[geezersoft]

What about using store virtual on the hosts?

[NB9457]

Interesting, this was sort of my original plan. I've got questions What do you use for VM backup? What is your total capacity after RAID for each server with those bad boys, that is if you are using RAID? How do you deal with a failed host and spinning them up on another host? Do you restore from backup, and if so, how long does it take? I know how longs a piece of string, but ball park per GB? What kind of SSD are you using? Enterprise stamped or just generic with good RAID

 

My main reason for a SAN is really survive-ability of VM's across hosts for host maintenance and uptime.

 

This kind of system running SSD's and mulitple 10GBe connections, even share nothing live migration could be a goer, I wonder what the drain time would be in that config!

 

Currently my game plan is 2 hosts and 1 san, but 3 hosts would take the san and it's proprietary parts out of the picture, cheaper support too.

 

hmmmmm

It's still in planning stages yet, but most of it worked out. We are intending to use DPM for VM backup, it's cheap, used it before and it's pretty good to be honest. VEEAM is extremely good, if you can afford it, I'd recommend it.

 

RAID we are looking at 8x960GB Enterprise Grade SSDs (either Samsung or Kingston). I was thinking RAID10 because I'm old school but RAID5 could well be an option.

 

As I said, we would re-purpose an old server to replica file services and SIMS, these are pretty much the core components to us and as long as these stay running, as far as staff are concerned, so is the network. So if a host had Student shares, WSUS, Classroom management, SCCM, Printers and Catering on and it failed, the Student Shares would be replicated, so flick a switch and it'll just appear as a hiccup. Then as we are using a VM infrastructure, we go to DPM, prioritise what needs to come back first, so maybe Catering then Printers, we simply copy the VHDs to another host and spin them up. As the VHDs for these services are no more than 60/80GB it takes only a few minutes to transfer the VHDs across the network. Broken host is then fixed within a day, services begin to get restored to it when it's back online (reverse replication and copy VHDs back onto it) and within 24-48hrs we are fully recovered and all staff will have noticed is maybe a few minute blip on the student shares and maybe 20 minutes without printers.

 

When you work out the money saved, the minor impact if a host fails and the likeliness of it even happening, it's something worth doing, especially seeing as you're not only simplifying the network, but getting better speeds due to faster storage etc. To be honest, it's more about having a good disaster plan then having the latest and greatest. We plan for every eventuality and what action we'd take in a given scenario, whether it be a fire or a drive failure. Heck, before I started, the place I'm at lost the SAN, as unlikely and as apparently redundant they are, it happened. Happened to a Uni in London too, lost pretty much everything.

[Steve21]

RAID we are looking at 8x960GB Enterprise Grade SSDs (either Samsung or Kingston). I was thinking RAID10 because I'm old school but RAID5 could well be an option.

 

I'm curious where you're going to find those servers with that spec + full SSDs for £6k~ (under the 3 for under £20k comment)

 

We're looking at sorting our DR out and a similar spec server without SSD is coming in more than that (Ok in all fairness it's online price so I know it'll come down after supplier discounts etc), but always good to see where people getting prices from

 

Steve

[NB9457]

I'm curious where you're going to find those servers with that spec + full SSDs for £6k~ (under the 3 for under £20k comment)

 

We're looking at sorting our DR out and a similar spec server without SSD is coming in more than that (Ok in all fairness it's online price so I know it'll come down after supplier discounts etc), but always good to see where people getting prices from

 

Steve

 

Depends whether we choose to supply our own drives. Kingston DC400 960GB drives are around £280exVAT each.

 

If we went with the drives from the system builder, it puts us almost bang on £20k. We are currently looking at Broadberry as a supplier

 

If refurbs aren't an issue, you could get drives for £6700 and servers for about £5000 and that would be an amazingly fast and cost effective solution.

[Steve21]

Depends whether we choose to supply our own drives. Kingston DC400 960GB drives are around £280exVAT each.

 

If we went with the drives from the system builder, it puts us almost bang on £20k. We are currently looking at Broadberry as a supplier

 

If refurbs aren't an issue, you could get drives for £6700 and servers for about £5000 and that would be an amazingly fast and cost effective solution.

 

Ah yeah that's a big saving on those ones. The samsungs we were looking at were nearer £500 a pop so another whack of change!

 

Will have to have a nosy at that option.

 

Thanks,

Steve

[NB9457]

Ah yeah that's a big saving on those ones. The samsungs we were looking at were nearer £500 a pop so another whack of change!

 

Will have to have a nosy at that option.

 

Thanks,

Steve

Think at Broadberyy, 2x8 Core, 128GB RAM, redundant PSU and 8x960GB SSDs comes in at around £6600. Personally compared with the big names, it seems a decent price.

[CAWJames]

Hmmm, interesting on that pricing, although I can't hit that price point on the configurator!

 

I am currently thinking, 2 hosts with 128GB RAM and around 8TB of usable storage after RAID, 5 should be fine, as the disadvantages of RAID 5 shouldn't be a problem with SSD's. Dual 10GBe, one for LAN, one for dedicated live migration/VM backup network.

 

Then a 3rd server with good old fashioned rust drives with enough capacity to hold the contents of both hosts to capacity using DPM, which we use already for our backups to home site and trickle backup to our offsite DPM server.

 

The simplicity is amazing lol but will fill all needs.