10 years ago you could count on about a 5 year IT planning cycle. Virtualization was interesting and the cloud was something in the sky. By the way 10 years ago was when VMware introduced it’s first server virtualization offering.5 years ago you probably thought 2 or 3 years was a good planning horizon and virtualization was a serious consideration.Today? 12 months is looking like forever.Virtual machines are taken up and ripped down in a matter of days. VM sprawl is something many are wrestling with. The cloud is has leapt out of the hype cycle and into many enterprise datacenters.What’s Next?
In the new dynamic data center, customers need more agility than ever and need to eliminate boundaries of individual storage systems.Workloads are unpredictable and cause resource planning constraintsTechnology refreshes are challenging. According to the TheInfoPro, they are the #3 challenge for storage adminsStorage and capacity efficiency are measured at the system level, not the data center level. Accordingly, Thin Provisioning buffer are limited to a single system.Storage Virtualization helps with the Workload unpredictability. But they are (A) expensive devices, (B) more infrastructure to manage and ( C) another point of failure. Further they reduce storage intelligence to its lowest common denominator.
Storage Federation is Distributed volume management across self-governing, homogenous systems using native peer-to-peer communication. In more common words, Storage Federation is easydata movement across a set of similar storage systems. And it is easy to use & non-disruptive.Customers can now manage resources at the data center level, not the system level.
Storage Federations is enabled by HP Peer Motion.HP Peer Motion enables customers to do 3 things:Workload Balancing – with today’s unpredictable workloads, customers now have the flexibility and capability to move workloads from overutilized assets to underutilized ones.Technology refreshes – Customers can non-disruptively add new storage to the infrastructure or migrate data from older systems to newer ones.Manage storage utilization and efficiency at the data center level, not the individual system level. For example, the same size thin provisioning buffer can be used for multiple systems, not just 1.HP Peer Motion is available from the smallest system, LeftHand VSA, running on a VM to the new HP P10000 3PAR.Peer Motion is available among HP LeftHand Storage Systems from VSA to P4800. it is also available on any 3PAR array running Inform 3.1.1 OS. Even 3PAR systems from 2002 can run Peer Motion if they upgrade to the right O/S. If they do not want to upgrade, they can still use Peer Motion, but only as a “Source”, not as a “Target”
The InfoPro had a survey for biggest storage challenges for IT admins. Tech refreshes came in at #3.Storage Federation enables Tech Refreshes to be easy and non-disruptive. In Storage Federation seminar #3, you will head about cloud service provider, Savvis, and how they use HP Peer Motion to address this challenge.HP Peer enables these tech refreshes & upgrades to be non-disruptive.Due to the ease of use, there is less planning and no risk.Further, with the ability to move workloads dynamically, you can actually extend the usable life of your infrastructure investments. As arrays age, simply move less demanding, less critical workloads to these older arrays. Nothing is as inexpensive as the storage you already own.Has this ever happened to you? Your storage array becomes over subscribed, either from a capacity perspective or a performance perspective or maybe both, despite your best intentions. You’ve reached the physical limitations of the array even though you’ve planned for headroom and thought through the load characteristics. But the profile of one or more workloads is not quite what you planned or has evolved over time. You’ve just out grown the physical size of the box. Despite the best planning, storage requirements don’t always grow exactly as forecast. So what can you do?Previously, you would have complex data migrations or buy more storage.Today, With the workload balancing capabilities of Storage Federation, you can non disruptively and transparently, load balance your data between the storage systems that support Storage Federation. You can think of Storage Federation doing for your data what VMware’s Vmotion does for your virtual servers – providing the flexibility and availability to put your resources where you need them, when you need them. Workload balancing with Storage Federation provides resource alignment and optimization because you’ll map workloads to the right storage. The flexibility to map workloads is something 3PAR customers are used to its mulit-tenant features like 3PAR Dynamic Optimization and Adaptive Optimization. These software features allow changing data’s quality of service at both the LUN and sub-LUN level. Now in data centers built on cloud and virtual solutions, flexibility is even more important. With Peer Motion, you achieve an even higher level of flexibility with the ability to move data across systems.For Workload Balancing – Peer Motion delivers:Improved Overall Performance – moving the right workloads to the right resources … easilyHotspot eliminationReduction in upfront planning – If you can quickly react to real-time workload changes, the need for the detailed upfront planning is reduced
Now, let’s talk about another use case for Peer Motion, that being Federated Thin Provisioning. We’ll start by first reviewing traditional thin provisioning that you likely use WITHIN an array. With modern scale-out storage architectures like LeftHand and 3PAR, Thin Provisioning is built-in to the architecture. With traditional scale-up arrays, in response to customer interest in Thin Provisioning, they bolted it on as a feature. According to IDC, roughly 75% of 3PAR customers use Thin Provisioning vs around 10% for traditional arrays. With Thin Provisioning, you can expect to drive up capacity utilization and lower costs associated with the storage in that array. But the amount of storage that you’ll provision is naturally capped based on the usable capacity of the array, shown by the red rectangle in this picture. You might over provision the INSTALLED capacity in the array, but you won’t provision more than the MAXIMUM INSTALLABLE capacity or the physical limit of the array, as demonstrated by the green rectangle. The physical limit is the boundary line that is safe, predictable and easy today. Typically there is some buffer too between the actual written data on an array and the capacity installed at any given moment. But that capacity buffer on an individual storage device is limited to just that device. So in this example above, you’ll notice that the buffer on the 3rd array (on the right) is approaching the maximum installed capacity of that array. This brings you to a decision point. Let’s assume that this array is 3 or 4 years old. Do you really want to invest more capacity in an aging platform?
Now imagine if you could pool that buffered capacity between storage systems. With Peer Motion and its ability to non-disruptively move data to another 3PAR Storage System, you can provision storage to an application beyond the installed capacity of a single array. You’ll be able to take better advantage of the power, cooling, acquisition and other cost savings ACROSS the data center, improving efficiency not only at the array level, but at the data center level. In this slide, you’ll see with Peer Motion, there is more opportunity to increase the capacity limit presented to the host, again the green rectangle, even though the maximum physical capacity of the array remains the same (the red rectangle). Now look closely at the 3rd array. You’ll see that its written data is close to or even at the maximum physical capacity of the array, maybe because its volumes are growing faster than predicted or because you have one runaway volume. Now you can “peer motion” that data to another array (shown on the far right), relieving the source array of its over provisioned state. In the mean time, the application is uninterrupted and unaware that the data has moved. Looking at this another way. Let’s say you have a 10TB Buffer for Thin Provisioning and 2 100 terabyte arrays. The maximum utilization if 90TB, with the 10TB buffer or 90%. With the flexibility of HP Peer Motion, you can manage that same 10TB across both arrays. Now, on average, the buffer is only 5TB/array, increasing potential utilization to 95%.Now can migrate to another array with out having to BUY more capacity, driving utilization to a much higher rate on installed arrays100 TB and 12 TB buffer x 3. 90% utilization4TB = 96% utilization
The other key here is that not only is Peer Motion non-disruptive, it is easy to use so customers can do it frequently. The more customers use it, the more value they derive from it.If Peer Motion required consulting services or complex planning, it would not be used often. That is not the case here.7x fewer steps than the competition. We achieve that because Peer Motion is built-in to the scale-out architecture of both LeftHand and 3PAR, it is not bolted-on to an existing, traditional storage architecture.
Traditional thinking & storage deployment does not work in today’s unpredictable worldStorage Federation enables customers to move workloads easily to respond to real-time demands and changes
Transcript of "What Is Storage Federation?"
What Is Storage Federation andWhat Problems Does It Solve?Sandeep SinghProduct Line Manager, HP StorageMay 2012
THE WORLD HAS CHANGED New Applications Compressed IT-as-a-Service Planning Horizon 105 89% of workloads virtualized by 2015 52Massive storage Planning cycle – a Massivedemand, unpredictable thing of the past consolidation, unpredictafuture ble workloads