Sunday, June 30, 2013
Network Management with datacenter abstraction layer (SCVMM 2012 R2)
This blog post will show some of the cool new stuff related to network virtualization, and especially the support of network virtualization gateways through standard based management with SCVMM 2012 R2 and Windows Server 2012 R2.
The Software Defined Datacenter story was alright, but not good with Windows Server 2012 and System Center 2012 SP1.
My personal take on that, was mostly because of the third party requirements for virtualization gateways. Cisco have been working on some stuff, and so have many others.
However, Microsoft has listened to the feedback from their partners and customers, and made this native in both products.
You can now have your own virtualization gateway running in a VM (Windows Server 2012 R2) and manage it, end-to-end with Virtual Machine Manager 2012 R2.
First of all: You must have a dedicated physical Hyper-V server for this in your fabric, which is hosting the virtual machines with the RRAS role installed.
This Hyper-V host should be considered as an edge server, and not joined to the domain.
The virtual machines hosting the RRAS role should be joined to the domain and can be highly available in a cluster, and this is quite critical for production environments.
If you have structured your host groups in VMM very well, it could look something like this:
Next, let us add the Network Virtualization Gateway to the fabric in VMM.
1. Navigate to the fabric pane in the VMM console, expand Network and right click Network Service to add a new network service
2. Give your network service a name and a proper description.
3. Specify manufacturer and model of the network service. Default, this is Microsoft and we must select the proper Model. You can see from the drop down list that you can add Microsoft Standards-Based Network switches, that will let you manage your switches and TOR switches. Microsoft Windows Server IP Address Management (IPAM) for a better integration with your entire Windows network infrastructure, and last but not least; Microsoft Windows Server Gateway.
4. Specify your Run As account that have permission on the VM to install the VMM agent and configure the network service
5. Specify the connection string. You can see the example in this step of the wizard. We need the VM host (in my case, it is TomWaits), and the RRASServer, which is the name of the virtual machine with the RRAS role installed. My RRAS server is NVGRE. Click next to proceed.
6. If the connection string would have included any ports for SSL, a certificate may have been required. In my case, this doesn’t apply.
7. Test and validate the network service configuration provider. This will run basic validation tests of the provider. Click test and verify that the critical tests are passed, and the others are implemented. Click next to proceed.
8. Specify the host groups for which the network service will be available. In my case, I want all of my hosts groups to have access to this service. Click next twice, and VMM will add the network service to fabric.
9. The last step that needs to be done, is to specify the configuration of each network connection on the virtualization gateway.
10. Go back to fabric, network service and right-click on your virtualization gateway to list the properties. Click on connectivity and select both front end connection and back end connection. We will dive more into this in the next blog post.
Hopefully, this blog post shown how easy it was to leverage the standard based management experience of network virtualization gateways with SCVMM 2012 R2.
My next blog post will focus more on network virtualization gateways, and how to create the service template for network virtualization gateways.
Thursday, June 27, 2013
Storage – with datacenter abstraction layer
I have recently blogged about the datacenter abstraction layer story in both Windows Server 2012 R2 and System Center 2012 R2.
In this blog post, I want to highlight some of the storage capabilities you are able to leverage through standard based management with Virtual Machine Manager 2012 R2.
Virtual Machine Manager takes care of the entire fabric.
A fabric may also be referred to as a stamp, or resource pools.
The fabric should contain the physical building blocks that make cloud computing possible for the organization, either if it is a private cloud or a service provider cloud.
Most organizations are handling their storage pools as single units, with no common management tools and must rely on manual documentation to see how everything is connected. I have ran into quite a few customers who are struggling to get an overview of their storage fabric in the context of virtualization and cloud computing.
This is where VMM comes to the rescue.
What VMM is able to do with your storage:
· End-to-end mapping
· Storage classification
· Allocation and Assignment
· Rapid provisioning
· SAN migration
· Array onboarding
· Storage Automation
· Scalable provisioning
· Extensive array support
· Standard based management
· File server management
· Scale-out server creation
· File share management
As you can see, this is massive part of the lifecycle management of your datacenter, and this blog post will show how you can leverage the iSCSI Target Server role in Windows Server 2012 R2.
First, we need a Windows Server 2012 R2 server with the File Server role installed, and especially the iSCSI Target Server role enabled.
In a nutshell, you must do the following on your Windows Server 2012 R2 server:
· Attach storage
· Create storage pools
· Create disks and volumes
· Create iSCSI disk
· Create iSCSI target
Once this is completed, we can start to do the magic within VMM.
Note: for Windows Server 2012 R2, it's no need to install the VMM storage provider on your Windows Server prior to adding it to VMM. Windows Server 2012 R2 supports this native.
1. Navigate to Fabric—>StorageàProviders.
Right click and add storage device. Here, you must select ‘SAN and NAS devices discovered and managed by a SMI-S provider’.
2. Make sure you select SMI-S WMI as the protocol, and specify your storage target, port and Run as account (since my storage server is within the same domain, I can use my domain account).
3. Once the discovery process has completed, you will find your storage device listed with the available capacity.
4. On the ‘Select Storage Device’ page, mark the storage you will manage with VMM and associate the proper classification (you can create classifications directly on this page.)
5. Click complete and you can follow the process in the jobs pane in VMM and see what’s discovered and detected by VMM. In this case, VMM detects the target and the initiators you have created on the storage server.
We have now discovered and added a storage device to VMM and can view if under ‘arrays’ in Fabric.
The next step is to allocate storage to your compute resources which is located in the host groups in Fabric.
1. Click on ‘Allocate Capacity’. This will fire up a new window where you can allocate both storage pools and LUNS to your host groups. I will add storage to my service provider cloud infrastructure host group.
2. Next, I’ll navigate to the properties of this host group and find the storage tab. From here, I can see the associated storage for this host group. Note that I can also allocate both storage pools and LUNs directly from here.
3. The next step will be to connect my hosts to the storage device. I will go to properties on one host at a time, click storage, and click ‘Add’. This will list the available storage connectors for you. In my case, iSCSI array is the available option.
4. Connect to your storage array from the drop down list and eventually configure advanced settings for the connection. Once this is done, VMM will create iSCSI sessions between target and initiator.
5. If we check the storage tab on the host once again, we can see my iSCSI array I just added. I can also create multiple sessions for MPIO.
This was a simple guide on how to add storage through standard based management with Virtual Machine Manager 2012 R2.
The next blog post will also show how to create clusters with VMM where we are using the available storage we added to VMM, to create disk witness and clustered shared volumes.
Sunday, June 23, 2013
After the announcements during North America’s version of TechEd, I was finally able to discuss datacenter abstraction layer with my customers, and explain how we are doing things in R2 of Virtual Machine Manager (System Center 2012 R2).
Let me take a few steps back and explain.
I work as a CTO at Lumagate and is also so honored to be one of the special MVP’s in the cloud and management space. My bread and butter is cloud computing, and I spend most of my time on tuning and playing with datacenter technologies like Hyper-V, Azure and Virtual Machine Manager.
Over the last years, I have visited hundreds of VMware customers and evangelized the hypervisor in Windows Server. My job got much easier with the release of Windows Server 2012 and SP1 for System Center 2012, and now it is about to become easier once again, with the upcoming R2 release of both products.
What do I mean with that?
I could always visit those customers again and show new features, talk a bit more about the Cloud OS and hybrid magic, but I am having better approach this time.
Datacenter abstraction layer (dal) is my friend, and VMM is the component I use to put everything into the context.
Instead of having the traditional comparison of hypervisors where we (eventually) always ends up discussing features, I am focusing on the bigger picture. The complete picture with every required datacenter component involved.
Hypervisors is nothing without networking and storage. If you have been working with Virtual Machine Manager in SP1, you must have ran into the networking discussions about software defined networking, logical switches, logical networks and so on. In other words, you better know what you are talking about and know the details of the OSI-model. The same applies for storage. Customers expect that you are familiar with storage technologies and capabilities, no matter what manufacturer they are using.
To be able to manage all of this from a single pane of glass, we need abstraction and integration.
This is where Microsoft is best and brightest.
Instead of having a similar approach as the others (competitors) where they buy some companies and creates 1:1 plug-ins, Microsoft is using standard based management that simplifies our life.
What to expect from System Center 2012 R2:
Standard-based management that delivers datacenter plug-n-play.
We have already seen the silhouette’s and shadows from this already in 2012 and 2012 SP1, where you can do cool stuff with your hardware through BMC protocols, SMI-S, SMP plus more, and more is about to come. No matter what network device or storage you are buying, you should just be able to plug it into your datacenter and use it with System Center. This is datacenter abstraction layer!
Microsoft is doing this on high volume components that will decrease the cost of them.
By having this approach, we can shift from planning and deploying to integrating systems to have more capacity and capability in our clouds.
Looking forward to a busy second half of this year, ready to demonstrate, deploy, sell and help customers around the globe with datacenter abstraction layer delivered by the Cloud OS.
My next blog post will be much more technical where we take a closer look on the actual technologies involved through WS-Man, WinRM and Powershell.
Thursday, June 6, 2013
TechEd is running this week in North America and we have scratched the surface of some huge upcoming releases.
I have paid closely attention to Windows Server 2012 R2 and System Center 2012 R2 content, and especially fabric related stuff like clustering, Hyper-V and Virtual Machine Manager.
VMM, which is my second home, is about to blow any boundaries it may have had previously.
Let us just stop and think for a second. A couple of years ago, we were managing our Hyper-V environment with SCVMM 2008 R2. What was the great benefits here?
· Creating and deploy Virtual Machines from VM templates to the proper hosts
· Integration with SCOM for PRO tips
· Live Migration and Quick Storage Migration management
· P2V and V2V
And that was about it.
There is no need to dig into the details, but SCVMM 2008 R2 did not have the best Windows Server integration we have seen. With that, I mean that cluster management, as well as general hypervisor management was poor. You would rather rely on Failover Cluster Manager to have an updated view of your configuration, since SCVMM 2008 R2 didn’t seem to keep up with it.
This has changed dramatically over the last years.
I have already covered the great enhancements of both SCVMM 2012 and SCVMM 2012 SP1 previously on this blog, and will therefor jump right ahead to the upcoming release, SCVMM 2012 R2.
This will be the component for your entire fabric infrastructure and cloud. You can quote me on that whenever you would like.
Let us just mention some of the features we can expect:
· Datacenter abstraction layer (dal) will be the key for integration with third parties, hardware and software. A standard based management API will make our lifes easier in the future, and SCVMM will give you an instant, consistent and consolidated view from one console
· Fabric management that covers “everything” related to storage and networking. It gives you end-to-end management of switches (TOR, gateways, IPAM, network virtualization, NIC teaming, QoS), storage (FC, SMB3.0, virtual SAN, virtual FC, deployment of storage, zoning, maskin, trim, mask, unmask, ODX support) and a lot more!
· VM templates on steroids = Service Templates. Service Templates will be your best friends when it comes to infrastructure deployment where the roles are living within a virtual machine. Deployment of the other System Center components through Service Templates, as well as much more flexibility in general.
· Integration with Windows Azure for Windows Azure Recovery Services. Hyper-V Replica plays a huge role in the next Win 2012 R2 release and can almost be synchronized (30 seconds). SCVMM 2012 R2 will enable DR for your cloud and at the VM level.
· And much, much more.
As you can see, SCVMM could easily be renamed to System Center Cloud Manager, but then we would have another SCCM J
There is a bunch to cover, and I will provide you with the guides on this blog as soon as I have my hands on the preview we are expecting.
Windows Azure Services for Windows Server has also evolved, and it is now called ‘Windows Azure Pack’ that truly leverage the capabilities in SCVMM/Windows Server 2012 R2 – Hyper-V from a service provide and tenant perspective. Expect this to be covered as well.
If you have any comments to what you would like to know more of, please leave a comment and I will respond.
Monday, June 3, 2013
I am visiting a bunch of customers over the year to discuss datacenter technologies and virtualization. You can bundle this together and then we have a nice cloud discussion.
However, there has always been a common challenge for those customers:
When I say storage, I can mention quite a few common challenges.
De-dupe, integration, backup, replication, cost, and tiering.
Yes. Tiering it is.
The datacenter has been missing this additional layer above the storage to support logical placement of hot data.
What does this mean?
You have some SSD disks, some SAS disks and som SATA disks in your preferred storage device.
You must manually place the right data on the right LUNs in order to make efficient use of your storage investment. However, things may change on the way during operations and heavily accessed data can become data that is not used that much anymore.
Have you ever heard about StorSimple? The company that Microsoft bought last year?
The technology that StorSimple provides is an appliance that you install in your datacenter and voila – you have cloud integrated storage that is able to move and balance your storage needs between SSD, SAS, SATA and Blobs in Windows Azure. All this so that you can get the most of your investment.
This particular technology will be available in Windows Server 2012 R2 as announced during the Keynote today, showed by Jeff Woulsey.
This will be a stunning feature where Windows Server 2012 R2 truly show how important it is as an infrastructure component in your private cloud. By providing all this through standards you can leverage Windows Server 2012 R2 completely with System Center 2012 R2.
Jeff did also show that de-dupe is now possible on volumes where running VMs live, and also provides much more IOPS compared to volumes where de-dupe is not enabled. Cool? I think so too.
More to come!