12. Emulate Hardware, Bypass Physical Limitations Shared “ Virtual Disk” x86/x64 physical or virtual machine with internal & external storage CPUs & RAM provide “mega caches” for ultra-high performance iSCSI & Fibre Channel SAN Virtual Storage Pool VM VM
13. Nodes Paired for High Availability (HA) Shared “ Virtual Disk” Active Active Synchronous Mirroring VM VM VM VM
14.
15. Manufacturer-Independent SAN-Wide Features Load Balancing Centralized Management High-speed Caching Async Remote Replication Thin Provisioning Virtual Disk Pooling RAID Striping Sync Mirroring (High-Availability) Online Snapshots Virtual Disk Migration Continuous Data Protection & Recovery
16. Unmatched SANs NON- STOP LIGHTNING-FAST WASTE-FREE At a fraction of the cost
17.
18.
19.
20.
21.
22. Radically Simple Virtualization Software Portfolio Radically Simple High Availability SANs for Virtual Servers & Virtual Desktops
23. Stretch for Metro-wide, Nonstop Access Shared “ Virtual Disk” Building 1 Building 2 Active Active VM VM VM VM Synchronous Mirroring over several Miles/Kilometers VM VM VM VM
25. Server & Storage Down > Switch to Hot Site Retrieve Shared “ Virtual Disk” Active Suspend Resume Building 1 Building 2 VM VM VM VM
26. Switch Back < Finish Maintenance @ Site 2 Shared “ Virtual Disk” Resume Suspend Retrieve Building 1 Building 2 VM VM VM VM
27.
28. Lowest Cost HA Option: Self-Contained Load Balance, Migrate & Fail-over VMs Using Internal Drives iSCSI SAN over Ethernet LAN Room A Room B Mirroring VM VM VM VM
29.
30. Testimonial Tom Radford, System Administrator, Washington Archives Management “ T echnology like this has previously been prohibitively expensive for a small to mid-size business customer. Sure, large companies have been deploying virtual servers and virtual SANs, but they have had to pay a huge price for doing so. DataCore's SANmelody enables the small business to get this kind of virtualization environment [and] made it possible to eliminate costly, storage related downtime.”
31. Business Continuity Made Practical Storage Virtualization Server Virtualization Server Virtualization Repurpose Existing Assets Building 1 Building 2 Server Virtualization DR Site Storage Virtualization Building 3 Central IT site Desktop Virtualization HA VM VM VM VM VM VM Remote Replication
32.
33.
34. Get the Most from Your Storage Yet, More Economical to Acquire, Manage, Protect & Maintain
Welcome to our webinar on Virtualization Imperatives. Whether you are new to virtualization or deep into it, there’s no question that storage factors heavily into every aspect of IT consolidation projects. How much so may not be readily apparent. After our discussion today you will be better able to anticipate those storage-related challenges and be well prepared to deal with them, even when money is at a premium. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
We’ll first look at how server and desktop virtualization initiatives influence central storage requirements with regards to availability, performance and utilization. This backdrop best reveals new implications for the design and rollout of your storage area network. It also uncovers operational and budgeting impacts that you may not have previously considered. You’ll then receive some very clear illustrations and guidelines for selecting between competing storage infrastructure proposals. Their major architectural and implementation differences will be illustrated, as we point out opportunities for cost reduction and long-term investment protection. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
While I’m sure that you are versed on the many virtues of server and desktop consolidation, I must alert you to those unexpected characteristics that could prove problematic if not attended to early in your planning. In some respects, life is simpler for IT before pursuing consolidation. Blasphemy, I know. The consequences of storage-related interruptions, for example, were much more isolated when each server had its own disks. You could coordinate planned downtime among a small set of users, often working on similar chores. Some of them could even continue to operate offline from their PCs or workstations pretty much unaffected during a few hour outage. [CLICK]
Now consider what happens when you take workloads from multiple underutilized systems and place them on a single physical server. Far more users are affected when you take down the physical machine to reconfigure, replace or upgrade components of your disk arrays. So coordinating a maintenance window becomes far more difficult. Not to mention how many more users will complain when disk bottlenecks arise from the aggravated workloads. But wasn’t server virtualization supposed to help IT get around these problems? Yes, but seemingly at the cost of expensive shared storage. Therein lays the rub. [CLICK]
To sidestep planned maintenance activities, the advanced software responsible for server virtualization can be instructed to move workloads between physical machines without downtime. Workload migration, as it’s called, entails suspending activity on one server and resuming it on another. Such migrations coupled with automated policies also allow an organization to recover quickly from unexpected failures or keep any one physical machine from getting overloaded. Under the covers, the server virtualization software regularly saves the state of its virtual machines on shared disks where another physical server can get to it. Thus the need for a robust, shared SAN. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
As physical servers are pooled into interchangeable processing units, stiffer demands are placed on the central storage devices. For one, direct-attached disks cannot be shared. Only SAN-capable devices will do. Clearly any new storage will need to be bigger, faster and more reliable than what was previously in place since we have that many more workloads dependent on it. Any clue how much that drives up project costs? [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
Take the case of this municipal police station. After doing a little homework, they expected to consolidate 40 application servers into a virtual infrastructure for around $200,000. [CLICK] Can you image their shock when the solution provider told them that the new 10 TB SAN alone would chew up $190,000 of their budget? Needless to say, this project came to a screeching halt. [CLICK] The customer then turned to a different solution provider. Rather than propose a wholesale replacement, they used DataCore Software to create a high-availability, shared storage pool from the customer’s existing equipment. The complete project was delivered within the original funding limits. In fact, the DataCore solution improved on the first proposal by mirroring the SAN to a co-location facility 10 miles away. Additional DataCore software licenses have since been configured to cover a distant contingency location to which they replicate critical volumes over the Internet. More on this shortly. [CLICK]
Suffice it to say that the police station is just one of numerous examples that we see every day- sometimes larger, sometimes smaller. And thousands more are being added each year. [CLICK]
These experiences teach us some very important lessons. Storage infrastructure accounts for the largest investment in most virtualization projects. Isn’t that startling? The resulting sticker shock causes many initiatives to stall. Some solution providers react by cutting corners in the initial rollout only to find that it later fails to address availability expectations and ultimately compromises IT operations. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
As our CEO points out, DataCore “ makes it practical for organizations large and small to remove physical, geographical and device-specific restrictions of data storage equipment that slow down, interrupt or endanger their computer operations.” By practical, he means straightforward and economically viable. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
At this point, there should be no doubt that Virtualization projects are composed of three inextricably-linked elements. Sure, servers get the most spotlight. While a fair share of attention is being put on centralizing desktop management. Yet the success of these initiatives ultimately hinges on the underlying storage area network. It is here that DataCore offers you unsurpassed advantages by using software to virtualize storage. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
We, like our server virtualization colleagues, emulate hardware with software, in the process bypass its many physical limitations. Rather than use a proprietary disk array with some embedded firmware, we run the emulation on standard Windows servers as portable code. The same proven software can run on today’s x86 servers and later be moved to faster, cheaper machines . No loss of investment and no change in your storage practice. Incidentally, the servers can be physical or virtual machines. The software exploits the large memories and incredibly powerful CPUs of these relatively inexpensive processors to cache and speed up read and write operations to disk. Incredibly, you can configure up to 1 Terabyte of SAN-wide cache per node to attain solid state disk performance without the added cost. The back-end storage may be internal drives and external arrays, existing or new. DataCore supports storage consumers running any of the popular operating systems including Windows, MacOS, Linux, NetWare, HP-UX, AIX, and Solaris. They connect through iSCSI or Fibre Channel even if the physical disk arrays are not capable of serving those environments directly. All the server and desktop virtualization products view us as well behaved, shared devices. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
DataCore servers are logically paired to achieve high availability configurations. Much in the way that a RAID device portrays many drives as one disk, DataCore’s software creates a highly available shared “virtual disk” by automatically mirroring all writes between nodes. More on these HA characteristic in a few slides. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
There are numerous advantages to architecting the storage pool in this way. For example, you can adapt the configuration to your budget constraints. In some organization, the choices may be driven by brand. In other, more price sensitive scenarios, you may opt to repurpose existing servers and storage hardware to arrive at the lowest cost. As with server virtualization, virtualizing storage provides hardware independence and much latitude in how to address individual needs. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
A comprehensive set of manufacturer-independent features come built in. They include virtual disk pooling, synchronous mirroring, high-speed caching, thin provisioning, online snapshots and remote replication to name a few. We can circle back later to describe individual functions as time permits. For now, the main point is that the same, rich, centrally managed services make any combination of storage devices under our control behave better than they do on their own, making the hardware choices more a matter of personal preferences or best value and less about their specific premium-priced capabilities.
This unique architecture, perfected over the past decade has proven to deliver non-stop, lightning-fast, waste-free SANs at a fraction of the cost of purpose-built hardware. [CLICK]
I’ll use the next few minutes to further explore storage in the virtual world. This closer look will familiarize you with critical aspects of our software solution and expose holes in how others attempt to address the storage needs of virtual infrastructures. I’ll start with larger scenarios and end with the smallest. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
Earlier, I put up this picture. On the surface, choosing one of the fast and so-called “bullet-proof”, high-end arrays seem like a reasonable technical solution. These arrays are internally redundant, equipped with dual storage controllers, fans, power supplies and RAID, and dual ported to connect independent paths between the servers and storage; All very important features to sidestep SAN fabric failures or disconnections. Seem like they would do the job nicely, doesn’t it? That, my friends, is a trick question! [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
Put aside the marketing brochures for a minute. What can cause this so-called bullet-proof array to go offline? And I don’t mean a bomb or a hurricane. How about a routine firmware upgrade? Best practices suggest that you take the entire enclosure down for this kind of periodic maintenance. The same goes before someone works inside the cage to prevent a technician from shorting the backplane with a screwdriver. Ever heard of people clipping power to both controllers accidentally? Not as rare as you think. Or the cases when a leaky pipe forced a storage system shutdown to prevent electrical shock? Point is that despite internally redundant components, the unit as a whole remains a single point of failure or disruption. Any time you have to take the central storage system down, multiple physical servers and hundreds of VMs will suffer downtime. Talk about scheduling nightmares? Certainly not the carefree IT environment that was promised. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
On paper, the solution seems straightforward. [CLICK] Cut the array down the middle. Then put each half in a different location with its own source of power, cooling and UPS. Split them far enough apart that the stray screwdriver or the water leak can only hurt one side while the other remains up. Stretch the mirror over a high-speed link. Then you could repair and upgrade each side independently. Unfortunately, that’s easier said than done. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
In our very real, mechanical world, the closest most vendors come to such configurations ends up raising the solution price substantially. Only the top of their product lines can handle replicating disks between separate arrays, which kicks in some very costly licenses, both for the embedded features and for the host-based software to coordinate failover. Probably a bit more than many of you can afford. And still, you would be disappointed. The two, fully replicated arrays cannot mimic one shared disk. So there goes the live migrations we desperately sought. It behaves more like an expensive Disaster Recovery solution requiring special scripts to remap drives before cutting over. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
DataCore, on the other hand, gives you a radically simple and practical approach to the high availability problem, without the chainsaw. It works with server and desktop virtualization products from VMware, Microsoft, Citrix, Parallels and Virtual Iron. The same goes for the rest of you non-virtualized servers. [CLICK]
No cutting involved here. Building on our earlier diagram, DataCore mirrors in lock step to the other server before signaling I/O completion. The servers can be several miles [35 kilometers] apart using a metropolitan area network, or they may be in the adjacent building; again, far enough to prevent any accident from harming both sides. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
When someone needs to attend to a DataCore server or the storage behind it, they can gracefully take it out of service. The client systems treat it as if a cable had been disconnected and switch to an alternate path, apparently to the same shared disk. In reality, this path connects them with the mirror image on the other DataCore server which remains up. The alternate DataCore server continues to field its original workloads as well. Knowing that its partner is temporarily down, it logs which blocks on the disks are being updated. Then when the partner comes back up, the two will resynchronize by transmitting the most recent information in those blocks. Again, neither the applications nor the users sensed that half of their storage infrastructure was intentionally turned off. There is no need to coordinate application downtime, since the users are not affected. Pretty powerful- yes? SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
Should both the application servers and storage pool need to been taken out of service, the system administrator will first request that the hypervisor migrate the VM workloads to the other server. Those VMs will temporarily rely on the other half of the storage pool for their needs. Unforeseen site-wide failures automatically trigger the same behavior from advanced hypervisors without having to make special calls to the DataCore nodes or invoke special agents. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
We can switch back to complete maintenance at site 2 at our leisure, never having to take down users. Once we complete those maintenance tasks, the environment will be back to full strength. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
This is it how it looks in the real world with application clusters and the DataCore high-availability SAN spread between two campuses, operating as one fully redundant configuration. Each side completely capable of running the hospital’s IT needs. [CLICK]
Now let’s collapse this for an entry level requirement. Many smaller companies can make due with a pair of servers to run their whole business. If these servers have a little headroom, then you can run DataCore on a guest OS as a virtual machine alongside other application VMs. Connect the two physical servers with a LAN and DataCore’s software will turn it into a virtual SAN using the internal or directly attached disks as the physical storage pool. Mirroring occurs transparently over the iSCSI SAN without any need for an separate new SAN. [CLICK] Conceptually, it is no different from the earlier scenario. Should the workloads grow to exhaust the server resources, you can simply move the DataCore licenses to dedicated physical servers. The software licenses are fully portable between servers. See what I mean by radically simple? [CLICK]
Work with your solution provider to choose from two software titles. SANmelody licenses pool up to 32 TBs behind them. Purchase a pair of SANmelody licenses to achieve high-availability between physically separated nodes as illustrated earlier. For larger SANs [CLICK] or one’s composed of high-end storage systems with intelligent multi-pathing front-end controllers of their own, pick SANsymphony software. It may be configured for N+1 redundancy to keep a highly available infrastructure even if one of the servers running the DataCore goes down. So if you need X DataCore servers to get the job done, you’d configure X + one more for extra protection. SANsymphony also offers a superset of advanced SAN-wide services as well as add-on options not available in SANmelody. SANsymphony can be used to pool a number of SANmelody virtual disk environments along with other storage. An important distinction that you are sure to appreciate: the manufacturer’s suggested retail price (MSRP) to step up from one DataCore software license to the next is simply the difference in their prices. This guarantees that the value of your license investment is never lost and carries forward when you expand functionality or capacity.
Tom Radford with Washington Archives Management makes a really good point. “ T echnology like this has previously been prohibitively expensive for a small to mid-size business customer. DataCore's SANmelody enables the small business to get this kind of virtualization environment [and] made it possible to eliminate costly, storage related downtime.” [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
Now let’s watch how easy it is to introduce DataCore software into a virtualized environment in steps, taking full advantage of the equipment that becomes free with server consolidation. [CLICK] First to address basic disk space pooling. Then to ensure high-availability. [CLICK] Followed by replication to a remote Disaster Recovery site. [CLICK] And as the capacity demands grow, you merely add more disks to the pool without interruption. By the way, with DataCore, you can address both your virtual and physical servers HA and DR requirements with a common solution. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
Here’s one customer running a fully redundant environment using HA pairs at each end and asynchronous remote replication between them for complete business continuity solution. As you might guess, they did it in stages. [CLICK]
The Advanced Site Recovery component of DataCore’s comprehensive business continuity portfolio allows central IT organizations to cost-effectively spread disaster recovery (DR) responsibilities among several smaller sites, allowing each of them to accept a more manageable role in keeping the business going. SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
You’ll be happy knowing that through DataCore’s software you’ll get the fullest use, the highest availability and fastest performance from your storage investments. You’ll also find it more economical to buy, provision, manage, protect and maintain your physical and virtual IT environment today and in the foreseeable future. [CLICK] SANsymphony Product Introduction Copyright 2000 DataCore Software Corporation
While we are a perfect fit in any of your virtualization initiatives, keep us in mind whenever you or your colleagues are helping making a storage move on physical machines. Be it adding a SAN, expanding capacity, upgrading disks, or developing backup and disaster recovery plans. [CLICK]