Cannot decide between 1U servers for growing company
-
@coliver said:
@scottalanmiller said:
@coliver said:
@ntoxicator said:
In my opinion. There would be more overhead
ISCSI LUN attached to Xen Hypervisor > VM > attached as local disk. Unless pass-through?
Slightly more overhead... probably an immeasurable amount. At the same time you are going against best practices and defeating many of the advantages of virtualization in one fell swoop by not attaching the storage to your hypervisor and presenting a virtual disk to the VM.
As I was writing out the downsides, I'm not actually sure that it is more overhead. Because the iSCSI has to be processed in software by the VM rather than in hardware by the host there is more network overhead in doing it to the guest.
Right, I agree with this I assumed that it would be slightly more processing overhead for the hypervisor but since it would be doing it anyway it wouldn't be anything additional.
Networking in the host is more efficient than in the guest. And both networking and storage is more efficient in Linux and Xen than in Windows. So double bonus on efficiency.
-
I've moved VM's on Citrix Xen Server and Storage to another LUN at the time (when installed the 2nd Synology)
It saturated the network.
The current SuperMicro 1U server only has 2 Intel NIC cards. I have them bonded via Xen Center and LACP enabled.
-
@coliver said:
@scottalanmiller said:
@coliver said:
@ntoxicator said:
In my opinion. There would be more overhead
ISCSI LUN attached to Xen Hypervisor > VM > attached as local disk. Unless pass-through?
Slightly more overhead... probably an immeasurable amount. At the same time you are going against best practices and defeating many of the advantages of virtualization in one fell swoop by not attaching the storage to your hypervisor and presenting a virtual disk to the VM.
As I was writing out the downsides, I'm not actually sure that it is more overhead. Because the iSCSI has to be processed in software by the VM rather than in hardware by the host there is more network overhead in doing it to the guest.
Right, I agree with this I assumed that it would be slightly more processing overhead for the hypervisor but since it would be doing it anyway it wouldn't be anything additional.
Guys, this is largely the reason we architected HC3 the way we did - give you the flexibility and HA of SAN/NAS without the complexity or overhead of VSA's and storage protocols. This is also why we built it specifically for the SMB and Mid-Market - at a price that makes sense specifically for our target market (not trying to sound too "salesy" but this is exacly why we built the platform).
-
Also my goal was to migrate to NFS storage away from iSCSI
as dealing with the RAW image or .cow2 image file is hell of alot easier.
-
@Aconboy said:
HC3
How long has HC3 scale been avail? Today is my first time hearing and being introduced with the option
-
@ntoxicator said:
@Aconboy said:
HC3
How long has HC3 scale been avail? Today is my first time hearding and being introduced with the option
We began it in 2008, and first customer ship was in late 2011. we have north of 5000 units in the field across 1800 or so customer sites.Take a look at www.scalecomputing.com
-
The Scale systems are excellent. I know NTG has one. I've worked with their systems a couple of years ago, and the performance was night & day VS VMware and a similarly sized SAN. And their systems work really well.
-
@ntoxicator said:
I've moved VM's on Citrix Xen Server and Storage to another LUN at the time (when installed the 2nd Synology)
It saturated the network.
The current SuperMicro 1U server only has 2 Intel NIC cards. I have them bonded via Xen Center and LACP enabled.
Yeah, storage migrations will do that
That's why you want block storage on a dedicated SAN if possible so that it uses its own "back channel" whenever possible so that it doesn't impact other things. -
@ntoxicator said:
Also my goal was to migrate to NFS storage away from iSCSI
as dealing with the RAW image or .cow2 image file is hell of alot easier.
Agreed and good plan
-
I had the Synology NAS setup as Block Level storage for the Volume that serves out the ISCSI Luns. mehhhhh.
the complications! Lol.
This is why I was wanting to move to all new design and setup being that I already essentially have data on a centralized setup.
Could I get away with dual Synology 12-bay NAS units? (running in HA/replication).Probably
I was thinking about having 10Gbe backbone/interconnect for the NAS + The VM Node servers. So all that traffic rides on the 10GBE backbone and would not touch the 1Gbe switches.
-
@ntoxicator said:
@Aconboy said:
HC3
How long has HC3 scale been avail? Today is my first time hearing and being introduced with the option
Quite some time, they aren't new. They were a storage vendor before moving into hyperconvergence but they are one of the leaders in the HC space. They've been around longer than the terminology
HC is just starting to hit its stride in the market, though, so you'll start hearing about Scale and their competitors more and more in the near future. -
@dafyre said:
The Scale systems are excellent. I know NTG has one. I've worked with their systems a couple of years ago, and the performance was night & day VS VMware and a similarly sized SAN. And their systems work really well.
Scale especially kicks butt for Windows performance because of their stack.
-
@ntoxicator - yup, we do that too - 10gig out of band from the LAN path for storage stack handling and cluster self-awareness.
-
@ntoxicator said:
Could I get away with dual Synology 12-bay NAS units? (running in HA/replication).
Well here is the issue there.... If you are doing this as a traditional NAS (NFS or SMB) for file shares like mapped Windows drives or automounted home directories for Linux users this works reliably and beautifully. The failover is smooth and transparent.
From shops that have tested this with virtualization, the failover is not fast enough and the VMs typically fail so that you don't get the failover that you are hoping for but instead an outage with potential of corruption. The failover is just not fast enough (at least in real world tested scenarios) to use with VM HA.
-
@Aconboy said:
@ntoxicator - yup, we do that too - 10gig out of band from the LAN path for storage stack handling and cluster self-awareness.
Yeah, we have a 10GigE fiber switching stack just to handle our OOB Scale communications! Talk about throughput capabilities!
-
So if anyone can explain to me
To do away with centralized storage such as what we have now and I've been moving to. I suppose this is what I've grown use to.
In order to have localized storage at the node/hypervisor level. one or many of the hypervisors would be storing all the data and sharing out the NFS? Then its replicated between? probably with DRBD Storage.
however, would would this be done with Citrix Xen Server for instance?
-
@ntoxicator said:
Essentially What I was looking to do was KVM / VM with complete HA.
I'm uncertain about keeping data local to individual servers. maybe because I have no experience with localized storage in an HA environment? Its all been shared centralized storage.
But you don't HA today. At least not at the storage level.
-
@ntoxicator said:
shit me for getting torn to shreds on here. Pissing contest.
Its much easier to verbalize than type out exact specifics.
What i meant by "I just seen as Windows iSCSI initiator working much better; more manageable and not limited."
Am I currently using windows iSCSI initator? NO
Do I wish I was using it: Yes?Why: Because I feel it would be easier to manage and connect an iSCSI LUN as localized storage and data storage. The larger 2TB storage holds all the windows network shares and user profile data.... thats the problem.
So disconnect the LUN from the XenServer and connect it directly to ProxMox, then give that drive to the windows VM. does that not work in ProxMox?
-
@ntoxicator said:
So if anyone can explain to me
To do away with centralized storage such as what we have now and I've been moving to. I suppose this is what I've grown use to.
In order to have localized storage at the node/hypervisor level. one or many of the hypervisors would be storing all the data and sharing out the NFS? Then its replicated between? probably with DRBD Storage.
however, would would this be done with Citrix Xen Server for instance?
there are a couple of methods used there in the hyperconverged space. Two different schools of thought emerged on how best to simplify the architecture while maintaining the benefits of virtualization.
-
Simply virtualize the SAN and it's controllers - also known as pulling the SAN into the servers. The VSA or Virtual San Appliance approach was developed to move the SAN up into the host servers through the use of a virtual machine. This did in fact simplify things like implementation and management by eliminating the separate SAN. However, it didn't do much to simplify the data path or regain efficiency. The VSA consumed significant host resources (CPU and RAM), still used storage protocols, and complicated the path to disk by turning the IO path from application->RAM->Disk into application->RAM->hypervisor->RAM->SAN controller VM->RAM->hypervisor->RAM->write-cache SSD->Disk.
-
Eliminate the dedicated servers, storage protocol overhead, resources consumed and associated gear by moving the hypervisor directly into the OS of the storage platform as a set of kernel modules, thereby simplifying the architecture dramatically while regaining the efficiency originally promised by Virtualization.
-
-
@ntoxicator said:
So if anyone can explain to me
To do away with centralized storage such as what we have now and I've been moving to. I suppose this is what I've grown use to.
In order to have localized storage at the node/hypervisor level. one or many of the hypervisors would be storing all the data and sharing out the NFS? Then its replicated between? probably with DRBD Storage.
however, would would this be done with Citrix Xen Server for instance?
You could also look at Starwinds Virtual SAN. Which could do this as well.
