How to size a Scale-out File Server

Hi,
We are looking to implement a 2-node 2012 R2 Scale-out File Server cluster (using SAS JBOD enclosure) for the primary purpose of storing the VHD files that will be accessed by a 4-node 2012 R2 Hyper-V cluster using 10 gigabit Ethernet (no RDMA). Our
environment can be characterised as having a large number of mostly idle VMs that experience sporadic, low intensity use (this is *not* a VDI environment). We have 2 questions.
1) To what extent is RAM a consideration for the SoFS servers? We can't find any documentation to suggest that there are benefits to be gained by having more RAM in the SoFS servers but we don't know if we should go with 8/16/32/64+ GB RAM in each
of the nodes.
2) With the need to keep costs down, we don't think RDMA / SMB-Direct NICs are going to be within our reach. Should we however look to have 2 * dual-port 10 Gbps NICs in both the SoFS & Hyper-V boxes?

Unless your VMs are read-intensive and you're going to deploy a CSV cache memory requirement for serving mostly idle VMs can be pretty low. However RAM is cheap these days so going for less then 16GB per node does not sound reasonable. Good sample, see:
Windows Server 2012 File Server Tip: Enable CSV Caching on Scale-Out File Server Clusters
http://blogs.technet.com/b/josebda/archive/2012/11/14/windows-server-2012-file-server-tip-enable-csv-caching-on-scale-out-file-server-clusters.aspx
How to Enable
CSV Cache
http://blogs.msdn.com/b/clustering/archive/2013/07/19/10286676.aspx
HYPER-V OVER SMB: SCALE-OUT FILE SERVER AND STORAGE SPACES
http://www.thomasmaurer.ch/2013/08/hyper-v-over-smb-scale-out-file-server-and-storage-spaces/
Hope this helped :)
StarWind VSAN [Virtual SAN] clusters Hyper-V without SAS, Fibre Channel, SMB 3.0 or iSCSI, uses Ethernet to mirror internally mounted SATA disks between hosts.

Similar Messages

How to backup a Scale Out File Server

Can anyone explain to me how i am supposed to backup my Scale Out File Server? It is used to host our VDI environment but we have thrown a few other minimal shares on there for stuff relevant to the VDI setup. Profile Disk/Scripts etc
Per this topic http://social.technet.microsoft.com/Forums/en-US/15561919-963b-4164-a307-e3696f5d21e1/adding-scaleout-file-servers-to-dpm2012-sp1?forum=dpmfilebackup
"DPM does not support protecting scale-out file servers (SOFS) for file shares.
I do not really know how to proceed. To make sure everything is backed up.

Hi,
Scale out file server support in DPM is for VM's (and SQL DB's I think), only and to be honest because of the metadata overhead using scale out file server and CSV as a general file server is not a good choice. To protect VM's running on a scale out
file server the VM's are backed up from the Hyper-V hosts but the Scale Out file servers do require a DPM agent on them. SMB 3.0 has a new feature which enables Remote VSS therefore the volumes (shares) are snapped from the Hyper-V hosts.
This document will give you the basic steps (http://technet.microsoft.com/en-us/library/hh757866.aspx)
hope this helps,
Regards,
Paul

Scale out file server client access point using public nic

Thoughts on this one.
I have a Scale Out File Server cluster with a Client Access Point. Whenever i talk to the Client Access Point it uses the public nics.
If i talk to the Scale Out File Server directly it uses the private like i want it to. How can i get the Client Access Point using the private nics?

Hi JustusIV,
Could you tell us why you want to modify the CAP use the “private” network, the CAP is used for client access, your clients may can’t access your cluster if modify your CAP
use private network, if you want know how to modify the CAP of a cluster you can refer the following KB:
Modify Network Settings for a Failover Cluster
http://technet.microsoft.com/en-us/library/cc725775.aspx
More information:
Understanding Access Points (Names and IP Addresses) in a Failover Cluster
http://technet.microsoft.com/en-us/library/cc732536.aspx
Windows Server 2008 Failover Clusters: Networking (Part 4)
http://blogs.technet.com/b/askcore/archive/2010/04/15/windows-server-2008-failover-clusters-networking-part-4.aspx
Hope this helps.
We
are trying to better understand customer views on social support experience, so your participation in this
interview project would be greatly appreciated if you have time.
Thanks for helping make community forums a great place.

Scale Out File Server for Applications using Shared VHDX

Just trying to get a definitive answer to the question of can we use a Shared VHDX in a SOFS Cluster which will be used to store VHDX files?
We have a 2012 R2 RDS Solution and store the User Profile Disks (UPD) on a SOFS Cluster that uses "traditional" storage from a SAN. We are planning on creating a new SOFS Cluster and wondered if we can use a shared VHDX instead of CSV as the storage
that will then be used to store the UPDs (one VHDX file per user).
Cheers for now
Russell

Just trying to get a definitive answer to the question of can we use a Shared VHDX in a SOFS Cluster which will be used to store VHDX files?
We have a 2012 R2 RDS Solution and store the User Profile Disks (UPD) on a SOFS Cluster that uses "traditional" storage from a SAN. We are planning on creating a new SOFS Cluster and wondered if we can use a shared VHDX instead of CSV as the storage that
will then be used to store the UPDs (one VHDX file per user).
Cheers for now
Russell
Sure you can do it. See:
Deploy a Guest Cluster Using a Shared Virtual Hard Disk
http://technet.microsoft.com/en-us/library/dn265980.aspx
Scenario 2: Hyper-V failover cluster using file-based storage in a separate Scale-Out File Server
This scenario uses Server Message Block (SMB) file-based storage as the location of the shared .vhdx files. You must deploy a Scale-Out File Server and create an SMB file share as the storage location. You also need a separate Hyper-V failover cluster.
The following table describes the physical host prerequisites.
Cluster Type
Requirements
Scale-Out File Server
At least two servers that are running Windows Server 2012 R2.
The servers must be members of the same Active Directory domain.
The servers must meet the requirements for failover clustering.
For more information, see Failover Clustering Hardware Requirements and Storage Options and Validate
Hardware for a Failover Cluster.
The servers must have access to block-level storage, which you can add as shared storage to the physical cluster. This storage can be iSCSI, Fibre Channel, SAS, or clustered storage spaces that use a set of shared SAS JBOD enclosures.
StarWind VSAN [Virtual SAN] clusters Hyper-V without SAS, Fibre Channel, SMB 3.0 or iSCSI, uses Ethernet to mirror internally mounted SATA disks between hosts.

Windows Server 2012 R2 Scale out file server cluster using server disks as CSV

Hi,
My question is if I can create a Scale Out File Server Cluster with CSV using the disks that comes with the servers, we have 2 servers with 2 arrays each one, 1 array for the OS files and 1 array that we could use for the CSV.
Regards.

Hi,
a SoFS needs some kind of shared Storage, this could be in the old days ISCSI or FC SAN and now also a Shared SAS JBOD with Clustered Storage Spaces.
If you have 2 Servers with "local" Disks you need some sort of Software to create a Shared Disk Layer out of that local Disks, like Starwind or DataCore.
Scale-Out File Server for Application Data Overview
http://technet.microsoft.com/en-us/library/hh831349.aspx
check out Step 1: Plan for Storage in Scale-Out File Server
oh i forgot the normal 4th Option, some Kind of Clustered Raid Controller like HP or Dell offer in some Solutions.
Udo
Udo, clustered RAID controllers still require SAS disks mounted into external enclosure *OR* everything mounted into single chassis for Cluster-In-A-Box scenario (but that's for OEMs). The only two benefits somebody would have with them are a) ability to
privide RAID LUs to Clustered Storage Spaces (non-clustered RAID controllers would be used as a SAS controllers only in pass-thru mode) and b) abillity to have caches synchronized so VM moved from one physical host to another would not start from the "cold"
state. Please see LSI Syncro reference manual for details:
Syncro 8i
http://www.lsi.com/downloads/Public/Syncro%20Shared%20Storage/docs/LSI_PB_SyncroCS_9271-8i.pdf
"LSI Syncro CS solutions are designed to provide continuous application uptime at a fraction
of the cost and complexity of traditional high availability solutions. Built on LSI MegaRAID
technology, the Syncro CS 9271-8i enables OEMs and system builders to use Syncro CS
controllers to build cost-effective two-node Cluster-in-a-Box (CiB) systems and deliver high
availability in a single self-contained unit.
Syncro 8e
http://www.lsi.com/downloads/Public/Syncro%20Shared%20Storage/docs/LSI_PB_SyncroCS_9286-8e.pdf
LSI Syncro CS solutions are designed to provide continuous application uptime at a fraction
of the cost and complexity of traditional high availability solutions. Built on LSI MegaRAID
technology, the Syncro CS 9286-8e solution allows a system administrator to build a costeffective,
easy to deploy and manage server failover cluster using volume servers and an offthe-
shelf JBOD. Syncro CS solutions bring shared storage and storage controller failover into
DAS environments, leveraging the low cost DAS infrastructure, simplicity, and performance
benefits. Controller to controller connectivity is provided through the high performance SAS
interface, providing the ability for resource load balancing, helping to ensure that applications
are using the most responsive server to boost performance and help prevent any one server
from being overburdened.
So... Making long story short: 8i is for OEMs and 8e is for end-users but require a JBOD.
Hope this helped :)
StarWind VSAN [Virtual SAN] clusters Hyper-V without SAS, Fibre Channel, SMB 3.0 or iSCSI, uses Ethernet to mirror internally mounted SATA disks between hosts.

Scale out SSAS server for better performance

HI
i have a sharepoint farm
running performance point service in a server where ANaylysis srver,reporting server installed
and we have anyalysis server dbs and cubes
and a wfe server where secure store service running
we have
1) application server + domain controller
2) two wfes
1) sql server sharepoint
1) SSAS server ( analysis server dbs+ reporting server)
here how i scaled out my SSAS server for better performance
adil

Just trying to get a definitive answer to the question of can we use a Shared VHDX in a SOFS Cluster which will be used to store VHDX files?
We have a 2012 R2 RDS Solution and store the User Profile Disks (UPD) on a SOFS Cluster that uses "traditional" storage from a SAN. We are planning on creating a new SOFS Cluster and wondered if we can use a shared VHDX instead of CSV as the storage that
will then be used to store the UPDs (one VHDX file per user).
Cheers for now
Russell
Sure you can do it. See:
Deploy a Guest Cluster Using a Shared Virtual Hard Disk
http://technet.microsoft.com/en-us/library/dn265980.aspx
Scenario 2: Hyper-V failover cluster using file-based storage in a separate Scale-Out File Server
This scenario uses Server Message Block (SMB) file-based storage as the location of the shared .vhdx files. You must deploy a Scale-Out File Server and create an SMB file share as the storage location. You also need a separate Hyper-V failover cluster.
The following table describes the physical host prerequisites.
Cluster Type
Requirements
Scale-Out File Server
At least two servers that are running Windows Server 2012 R2.
The servers must be members of the same Active Directory domain.
The servers must meet the requirements for failover clustering.
For more information, see Failover Clustering Hardware Requirements and Storage Options and Validate
Hardware for a Failover Cluster.
The servers must have access to block-level storage, which you can add as shared storage to the physical cluster. This storage can be iSCSI, Fibre Channel, SAS, or clustered storage spaces that use a set of shared SAS JBOD enclosures.
StarWind VSAN [Virtual SAN] clusters Hyper-V without SAS, Fibre Channel, SMB 3.0 or iSCSI, uses Ethernet to mirror internally mounted SATA disks between hosts.

Scale out sharepoint server search to index server and Microsoft SharePoint Foundation Web Application to new server

Hi
in my single server these below service are running:
SharePoint Server Search
User Profile Service
Microsoft SharePoint Foundation Web Application
here how i scale out Microsoft SharePoint Foundation Web Application to another server
and SharePoint Server Search to new index server.
adil

Check here to see how to add servers to the farm:
http://technet.microsoft.com/en-us/library/cc261752(v=office.15).aspx
What Inderjeet meant was that if you chose to install SharePoint as a single server instead of a complete farm, you won't be able to add servers to the farm at a later time. You can check if this is the case by checking the registry key he mentioned. If
that is indeed the case, there's nothing to it but reinstall the farm.
So that is why you never should choose the single server option, it locks you in and limits future options, while choosing to install sharepoint as a complete farm it's still perfectly acceptable to host it on a single server.
Kind regards,
Margriet Bruggeman
Lois & Clark IT Services
web site: http://www.loisandclark.eu
blog: http://www.sharepointdragons.com

How to convert concurrent program out files .out file to .txt files in

Hi
Trying to know if there is a way to convert the concurrent programs' output files witn '.out' extension to files with extension '.txt'
thanks
kp

Why you want to change the file extension to txt instead of out?
I believe you cannot do and you are not supposed to change it. Controlling the report output to different types can be done as explained in these docs.
How to Setup The Report Output to Different Viewer Types in Oracle Applications 11i [ID 184375.1]
How to Control the Name and the Application that Opens the Concurrent Request Output File on the Client? [ID 316752.1]
Thanks,
Hussein

How do I truly wipe out files from my Mac?

I have a ton of music and videos on itunes. I often right click on the file and select move to trash. Is this all I have to do to truly clear out the file from my computer? Or do I need to go into the finder and manually grab the file and toss it in the trash?
Thanks!
Jay.

"move to trash" really does move the file to the trash. once you empty the trash, the file is gone.

Restoring new DB to SMB3 Scale Out Server Disk takes forever to initialize

We have a SQL Server 2014 DB installed on a VM running 2012 R2 Standard with Datacenter 2012 R2 as the host. We have a Scale Out File Server (SOFS) running 2012 R2 on two nodes clustered (CSV). The entire network is 10GB. SQL server is configured
to use the SOFS cluster for it's databases files (using SMB3 as a network share). When restoring a backup to the server as a new database restore, the process takes about an hour (it's a big DB 210GB). When restoring the backup over the top
of the existing database (nearly similar file/DB sizes), the restore takes 10 minutes. I have read about "Instant File Initialization" but that appears to only apply to directly attached disks where the OS can block allocate the data using
some "Volume Maintenance" permissions. The network traffic during the "initialization phase" seems to be floating about 700mbs, yet when the actual restore counter starts (reading the backup file and writing the data) the network
activity steps up to 6.5Gbs - sometimes higher - the count down appears to take 10 minutes either way. So I'm quite sure that the 50 minutes is attributable to the initialization phase. Which seems horribly excessive.
Has anyone had similar experience using this kind of configuration?

For these databases, we use the simple recovery model. So the log file is not very large (5GB compared to the 200GB of the data file).
It appears that it is spending it's time zeroing the MDF file:
11/3/2014 5:38
spid55
Unknown
SetTargetRestoreAge: 0
11/3/2014 5:38
spid55
Unknown
Restore: LogZero complete
11/3/2014 5:38
spid55
Unknown
Restore: Waiting for log zero on FOO_TEST
11/3/2014 5:28
spid55
Unknown
Restore: Transferring data to FOO_TEST
11/3/2014 5:28
spid55
Unknown
Restore: Restoring backup set
11/3/2014 5:28
spid55
Unknown
Restore: Containers are ready
11/3/2014 5:28
spid55
Unknown
Zeroing completed on \\SOFSC\Shares\DB\FOO-DBD1\FOO_TEST.mdf
11/3/2014 4:31
spid55
Unknown
Zeroing completed on \\SOFSC\Shares\DB\FOO-DBD1\FOO_TEST_log.ldf
11/3/2014 4:30
spid55
Unknown
Zeroing \\SOFSC\Shares\DB\FOO-DBD1\FOO_TEST.mdf from page 1 to 26289280 (0x2000 to 0x3224900000)
11/3/2014 4:30
spid55
Unknown
Zeroing \\SOFSC\Shares\DB\FOO-DBD1\FOO_TEST_log.ldf from page 1 to 714448 (0x2000 to 0x15cda0000)
11/3/2014 4:30
spid55
Unknown
Restore: PreparingContainers
So it looks like the "Instant File Initialization" settings do not apply to shared disk resources?

Which role do I need DFS or File server on fail over cluster server 2012 R2?

what I want to achieve is that I want to share all my user data files in a central location and to be highly available all the time whether it's a general share or folder redirection data. BUT I'm a bit confused; I have fail over cluster set-up
on server 2012, now I would like to add DFS as a role but than we have another role called File server and virtually it does the same thing as DFS? Means it creates a namespace share that can be access even one of the nodes goes down. Now I am thinking is
that DFS does the replication between two physical location but fail over cluster works slightly differently and with file server it pretty much does the same thing except for replicating data from one drive to another. Now what do you suggest I do or
did I get the concept wrong like a noob?

DFS and Failover Clustering for file shares provides a similar end result for file access, but they are significantly different implementations.
Clustering provides high availability to files by presenting shared access to set a files served from a cluster. With 2012 R2 Microsoft added the ability to create a Scale-out File Server that even allows all nodes of the cluster to server access to
the files for a higher level of performance and other great things. Bottom line with Failover Clusters for files is that there is a single copy of the file presented from the cluster.
DFS on the other hand provides high availability to files by presenting multiple copies of the file by making a copy in two or more locations and presenting a naming space that allows access to the file through any of the network paths. DFS works very
well for files that are primarily read-only. When you get into a situation where there is a lot of updating of the shared files, DFS is not a very good solution. There are ways to implement DFS for read/write files, but it generally requires a
good knowledge of how the files are used and how you want to manage them.
The key to answering your question comes in your first sentence "I want to share all my user data files in a central location and to be highly available all the time". My initial reaction to this is that central location means Failover Cluster
- there is only a single copy of the file. However, "all the time" can be compromised by network failures to the central site. Remote sites would not have access if they can't access the central site. DFS provides the ability to
have copies remotely, but then if you allow updating at multiple sites, you have to manage the merging of the changes, among other things.
. : | : . : | : . tim

File Server Failover Cluster without shared disks

i have two servers that i wish to cluster as my Hyper V hosts and also two file servers each with 10 4TB SATA disks, all i have read about implementing high availability at the storage level involves clustering the file servers (e.g SOFS) which require external
shared storage that the servers in the cluster can access directly. i do not have external storage and do not have budget for it.
Is it possible to implement some for of HA with windows server 2012 R2 file servers without shared storage? for example is it possible to cluster the servers and have data on one server mirrored real-time to the other server such that if one server goes
down, the other server will take over processing storage request using the mirrored data?
i intend to use the storage to host VMs for Hyper V fail-over cluster and an SQL server cluster. they will access the shared on the file server through SMB
each file server also has 144GB SSD, how can i use it to improve performance?

i have two servers that i wish to cluster as my Hyper V hosts and also two file servers each with 10 4TB SATA disks, all i have read about implementing high availability at the storage level involves clustering the file servers (e.g SOFS) which require external
shared storage that the servers in the cluster can access directly. i do not have external storage and do not have budget for it.
Is it possible to implement some for of HA with windows server 2012 R2 file servers without shared storage? for example is it possible to cluster the servers and have data on one server mirrored real-time to the other server such that if one server goes
down, the other server will take over processing storage request using the mirrored data?
i intend to use the storage to host VMs for Hyper V fail-over cluster and an SQL server cluster. they will access the shared on the file server through SMB
each file server also has 144GB SSD, how can i use it to improve performance?
There are two ways for you to go:
1) Build a cluster w/o shared storage using MSFT upcoming version of Windows (yes, finally they have that feature and tons of other cool stuff). We've recently build both Scale-Out File Server serving Hyper-V cluster and standard general-purpose File Server
cluster with this version. I'll blog next week edited content (you can drop me a message to get drafts right now) or you can use Dave's blog who was the first one I know who build it and posted, see :
Windows Server Technical Preview (Storage Replica)
http://clusteringformeremortals.com
Feature you should be interested in it Storage Replica. Official guide is here:
Storage Replica Guide
http://blogs.technet.com/b/filecab/archive/2014/10/07/storage-replica-guide-released-for-windows-server-technical-preview.aspx
Will do things like on the picture below:
Just be aware: feature is new, build is preview (not even beta) so failover does not happen transparently (even with CA feature of SMB 3.0 enabled). However I think tuning timeout and improving I/O performance will fix that. SoFS failover is transparent
even right away.
2) If you cannot wait 9-12 months from now (hope MSFT is not going to delay their release) and you're not happy with a very basic functionality MSFT had put there (active-passive design, no RAM cache, requirement for separated storage, system/boot and dedicated
log disks where SSD is assumed) you can get advanced stuff with a third-party software doing things similar to the picture below:
So it will basically "mirror" some part of your storage (can be even directly-accessed file on your only system/boot disk) between hypervisor or just Windows hosts creating fault-tolerant and distributed SAN volume with optimal SMB3/NFS shares.
For more details see:
StarWind Virtual SAN
http://www.starwindsoftware.com/starwind-virtual-san/
There are other guys who do similar things but as you want file server (no virtualization?) most of them who are Linux/FreeBSD/Solaris-based and VM-running are out and you need to check for native Windows implementations. Guess SteelEye DataKeeper (that's
Dave who blogged about Storage Replica File Server) and DataCore.
Good luck :)
StarWind VSAN [Virtual SAN] clusters Hyper-V without SAS, Fibre Channel, SMB 3.0 or iSCSI, uses Ethernet to mirror internally mounted SATA disks between hosts.

IIS Web Farm - File Server for Content and Shared Configuration - SOFS?

We currently have a number of web farms with their content and shared configuration files located on a standalone file server.
I am looking to utilise the clustered file server role in Windows 2012 to provide improved uptime and load balancing for the configuration and content shares.
Please could someone provide clarification as to whether SOFS is supported for this scenario?
There appears to be some conflicting advice in the documentation and across the internet.

We currently have a number of web farms with their content and shared configuration files located on a standalone file server.
I am looking to utilise the clustered file server role in Windows 2012 to provide improved uptime and load balancing for the configuration and content shares.
Please could someone provide clarification as to whether SOFS is supported for this scenario?
There appears to be some conflicting advice in the documentation and across the internet.
SoFS is a good choice for a) Hyper-V and b) SQL Server (anything working with a big files like VHDX and MDF). For a bunch of a small files IIS has in a shared directories SoFS is not going to work and it's cleary said in the documentation. See:
Scale-Out File Server for Application Data Overview
http://technet.microsoft.com/en-us/library/hh831349.aspx
Workload: Information worker
Yes
Not recommended
Workload: Hyper-V
Yes
Yes
Workload: Microsoft SQL Server
Yes
Yes
So in your case just an ordinary clustered file servers should be enough. See:
Create a Clustered File Server
http://technet.microsoft.com/en-us/library/cc753969.aspx
Hope this helped :)
StarWind VSAN [Virtual SAN] clusters Hyper-V without SAS, Fibre Channel, SMB 3.0 or iSCSI, uses Ethernet to mirror internally mounted SATA disks between hosts.

Migrate User Profile Disks to new File Server

I'm looking for some guidance/best practice/advice on moving my user profile disk share. My goal is simple enough, we have a share with UPDs on a stand-alone file server. I'm in the process of configuring a scale-out file server and would like to migrate
my UPD share to that cluster once it is running smoothly. I'm running a 2012 R2 RDS farm. Is the process as simple as changing the UNC path in Server Manager and migrating the data? If so, should I copy the data then change the UNC path or the other way around?
On a related note, is there any problems of hosting the UPD share on DFS? I'm not necessarily looking to replicate the share, but it would ease a similar migration in the future if necessary as I could replicate it or use the DFS migration toolkit without
needing to alter the UNC path in the farm as the client facing path wouldn't change. Just an idea I'm toying with and would love some feedback.

I would think in a simple/flat scenario, you change the UNC path and the wizard will automatically setup the share on the target folder with the correct root permissions
then you can do a robocopy.exe /mir to move the individual UPDs over while retaining their unique permissions for each user
regarding scale out file server, you actually can't use SoFS for just a general file share workload and SoFS + DFS combination also has limited capabilities
for now, SoFS is only compatible with clustered Hyper-V, IIS, VMM and SQL workloads
you can use the old fashion clustered file server role with SMB 3.0 for the UPDs however, as 2012 now allows for proper clustered share aliases that allows you to natively use a virtual name under a clustered file share option which works nicely in your
case
setting up DFS with the above scenario seems to be supported although I haven't done it before
http://technet.microsoft.com/en-us/library/hh831349.aspx

High availability for file server and exchange with 2 physical servers

Dear Experts,
I have 2 physical server with local disks only. I want to setup below on same with high availability, please advise best prossible options. We will be using windows 2012 R2 Server..
1. Domain controller
2. Exchange 2013
As of now I am thinking of setting up below:
1. Install Hyper-v on both and create 3 VM on each as
-On Host A- 1 VM for DC, 1 VM for File server with DFS namespace and replication for file server HA and 1 VM for Exchange 2013 with CAS/MBX with DAG and DNS RR for Exchange HA
-On Host B - 1 VM for ADC, 1 VM for File server DFS member for above and 1 VM for Exchange 2013 CAS/MBX with DAG member
I have read on internet about new features called scale out file server (SoFS) in Windows 2012 Server but not sure that will be preferred for file sharing.
Any advise will be highly appreciated..
Thanks for the help in advance..
Best regards,

Dear Experts,
I have 2 physical server with local disks only. I want to setup below on same with high availability, please advise best prossible options. We will be using windows 2012 R2 Server..
1. Domain controller
2. Exchange 2013
As of now I am thinking of setting up below:
1. Install Hyper-v on both and create 3 VM on each as
-On Host A- 1 VM for DC, 1 VM for File server with DFS namespace and replication for file server HA and 1 VM for Exchange 2013 with CAS/MBX with DAG and DNS RR for Exchange HA
-On Host B - 1 VM for ADC, 1 VM for File server DFS member for above and 1 VM for Exchange 2013 CAS/MBX with DAG member
I have read on internet about new features called scale out file server (SoFS) in Windows 2012 Server but not sure that will be preferred for file sharing.
Any advise will be highly appreciated..
Thanks for the help in advance..
Best regards,
DFS is by far the best way to implement any sort of file server. Because a) failover is not fully transparent and does not happen always (say not on copy ) b) DFS cannot replicate open files so if you edit a big file and have node rebooted you're going to
lose ALL transactions/updates you've applied c) actually slows down the config. See:
DFS for failover
http://help.globalscape.com/help/wafs3/using_microsoft_dfs_for_failover.htm
DFS FAQ
http://technet.microsoft.com/library/cc773238(WS.10).aspx
(check "open files" point here)
DFS Performance
http://blogs.technet.com/b/filecab/archive/2009/08/22/windows-server-dfs-namespaces-performance-and-scalability.aspx
SoFS a) requires shared storage to run and you don't have one b) does not support generic workloads
(only Hyper-V and SQL Server) and c) technically makes sense to expand SAS JBOD or existing FC SAN to numerous Hyper-V clients over 10 GbE w/o need to invest money into SAS switches and HBAs and FC HBAs and new licenses FC ports. Making long story short:
SoFS is NOT YOUR CASE.
SoFS Overview
http://technet.microsoft.com/en-us/library/hh831349.aspx
http://www.aidanfinn.com/?p=12786
http://www.aidanfinn.com/?p=12786
For now you need to find some shared storage to be a back end for your hypevisor config (SAS JBOD from supported list, virtual SAN from multiple vendors like for example StarWind see below, make sure you review ALL the vendors) and then you create a failover
SMB 3.0 share for your file server workload. See:
Clustered Storage Spaces over SAS JBOD
http://technet.microsoft.com/en-us/library/jj822937.aspx
Virtual SAN from inexpensive SATA and no SAS or FC
http://www.starwindsoftware.com/native-san-for-hyper-v-free-edition
Failover
SMB File Server in Windows Server 2012 R2
http://www.starwindsoftware.com/configuring-ha-file-server-on-windows-server-2012-for-smb-nas
Fault
tolerant file server on just a pair of nodes
http://www.starwindsoftware.com/ns-configuring-ha-file-server-for-smb-nas
For Exchange you use SMB share from above for a file share witness and use DAG. See:
Exchange DAG
Good luck! Hope this helped :)
StarWind VSAN [Virtual SAN] clusters Hyper-V without SAS, Fibre Channel, SMB 3.0 or iSCSI, uses Ethernet to mirror internally mounted SATA disks between hosts.

How to size a Scale-out File Server

Similar Messages

Maybe you are looking for