Delivering on the SAN Promise

2. Instructions: You will typically arrive here by jumping from slide 5. Based on the profile you�ve described, it looks like a few of your key issues would be well addressed by: Organizing � All of this is possible with DataCore�s Advanced Cross-Platform Storage Virtualization offering.Instructions: You will typically arrive here by jumping from slide 5. Based on the profile you�ve described, it looks like a few of your key issues would be well addressed by: Organizing � All of this is possible with DataCore�s Advanced Cross-Platform Storage Virtualization offering.

3. How Softek Virtualization Works Softek Virtualization solutions allow arbitrarily sized virtual disks from a mixed storage pool to be un-intrusively added, moved and replicated on behalf of SAN-connected open systems without host modifications using a secure centralized GUI. Advanced SAN-wide services including caching, snapshots and remote mirroring enhance the properties of the underlying storage devices for improved availability, performance and uniformity. The result is a scalable network storage pool comprised of virtual disks carved from a mix and match of JBODS (just a bunch of disks), disk arrays and conventional storage subsystems, that is available to the popular open system OS flavors, independent of vendor. Perhaps this drawing will better describe what is going on. Several application servers running the popular operating systems are connected into a Fibre Channel switch. So too are various storage devices from well known suppliers. A storage domain server (SDS) is added into the configuration to create and manage the network storage pool. This simple configuration is intended to illustrate the role of each element. More sophisticated networks would be configured to ensure highly available access to the storage resources. Note that the storage devices have been zoned to the storage domain server, as have the application server. The servers do not see the actual physical disks. They will only see the logical volumes (not shown) allocated to them by the storage domain server. To the application servers, the logical volumes appear as directly attached disks. Existing SCSI, EIDE and SSA storage devices may be included in the pool by attaching them directly to storage domain servers. To the hosts, virtual disks served from these devices appear as Fibre Channel volumes. Note: Re-commissioning the legacy arrays is typically done after their data is cleansed and the space is repartitioned. Softek Virtualization solutions allow arbitrarily sized virtual disks from a mixed storage pool to be un-intrusively added, moved and replicated on behalf of SAN-connected open systems without host modifications using a secure centralized GUI. Advanced SAN-wide services including caching, snapshots and remote mirroring enhance the properties of the underlying storage devices for improved availability, performance and uniformity. The result is a scalable network storage pool comprised of virtual disks carved from a mix and match of JBODS (just a bunch of disks), disk arrays and conventional storage subsystems, that is available to the popular open system OS flavors, independent of vendor. Perhaps this drawing will better describe what is going on. Several application servers running the popular operating systems are connected into a Fibre Channel switch. So too are various storage devices from well known suppliers. A storage domain server (SDS) is added into the configuration to create and manage the network storage pool. This simple configuration is intended to illustrate the role of each element. More sophisticated networks would be configured to ensure highly available access to the storage resources. Note that the storage devices have been zoned to the storage domain server, as have the application server. The servers do not see the actual physical disks. They will only see the logical volumes (not shown) allocated to them by the storage domain server. To the application servers, the logical volumes appear as directly attached disks. Existing SCSI, EIDE and SSA storage devices may be included in the pool by attaching them directly to storage domain servers. To the hosts, virtual disks served from these devices appear as Fibre Channel volumes. Note: Re-commissioning the legacy arrays is typically done after their data is cleansed and the space is repartitioned.

4. Centralizes Storage Management Controls and monitors the virtualized storage pool under a consolidated, intuitive, graphical interface Centralize mapping of virtual volumes, replication and business continuance All storage control is securely centralized under the jurisdiction of a SAN-wide storage administrator, who implements the storage policy according to business priorities established for each site. Application server is seen as a local disc One GUI to view all Mirroring, replication, snapshot, off site DR . No need to use three different interfaces.. All storage control is securely centralized under the jurisdiction of a SAN-wide storage administrator, who implements the storage policy according to business priorities established for each site. Application server is seen as a local disc One GUI to view all Mirroring, replication, snapshot, off site DR . No need to use three different interfaces..

5. Virtualization Simplifies Provisioning to a Drag-and-Drop

6. Centralize Business Continuance and Disaster Recovery in One Tool N+1 Availability Alternate pathing Mirroring Point in Time Replication Snapshot Asynchronous on-site and off-site IP Mirroring Linear scaling Hardware based mirroring in the box is expensive. Virtualisation eliminates single point of failure by allowing mirror between boxes. Expensive storage for mission critical, cheap stuf for snapshots, etc. Replication - traditionaly single vendor, Virualisation gives any vendor replication on or off site. Snapshot point in time copy. Cost of replication products can cost more than virtualisation. Back up can go from off site - volume Archive data asyncronous IP mirror gets rid of back up window. Each SDS can aim off site. N+1 redundancy of switches, storage, and SDSs. Fault tolerant duplexing Alternate pathing circumvents the path, node, and device failures in the SAN Alternative to host-based mirroring Rapidly restores an application server by booting from virtual disks Backup window is no longer necessary, instantaneous point-in-time copies without disruptions Instantaneously creates point-in-time images of online data Asynchronous and synchronous mirroring between multiple sites connected by LANs, MANs, and WANs, using standard TCP/IP and FC protocols Hardware based mirroring in the box is expensive. Virtualisation eliminates single point of failure by allowing mirror between boxes. Expensive storage for mission critical, cheap stuf for snapshots, etc. Replication - traditionaly single vendor, Virualisation gives any vendor replication on or off site. Snapshot point in time copy. Cost of replication products can cost more than virtualisation. Back up can go from off site - volume Archive data asyncronous IP mirror gets rid of back up window. Each SDS can aim off site. N+1 redundancy of switches, storage, and SDSs. Fault tolerant duplexing Alternate pathing circumvents the path, node, and device failures in the SAN Alternative to host-based mirroring Rapidly restores an application server by booting from virtual disks Backup window is no longer necessary, instantaneous point-in-time copies without disruptions Instantaneously creates point-in-time images of online data Asynchronous and synchronous mirroring between multiple sites connected by LANs, MANs, and WANs, using standard TCP/IP and FC protocols

7. Dramatic Asset Utilization Improvements 30% � 50% 300 � 600 GB 80% + 8,000 � 15,000 GB Double disk utilization by tapping into misallocated or unreachable capacity Free space fully accessible across network of hosts No longer tied to specific server Allocation units (GB/ Volume) match requirements Independent of physical spindle size Performance of disks (even unintelligent arrays) accelerated by network-wide caching Offloads much of disk read-access Makes existing disks suitable for more applications Higher application throughput and SAN performance Storage-related tasks offloaded from application hosts More responsive applications More responsive production LANs Reduction in planned downtime means More time available for online processing More time available revenue generating workloads Improved staff productivity/coverage Achieve 25X greater coverage through consolidation of distributed storage assets Manage more with current staff Fewer people required to manage storage Centralized management Intuitive drag-and-drop GUI Selective privileges Enhanced security Only explicitly assigned virtual volumes are visible to hosts Rogue hosts are prevented from accessing or discovering neighbor�s resources Double disk utilization by tapping into misallocated or unreachable capacity Free space fully accessible across network of hosts No longer tied to specific server Allocation units (GB/ Volume) match requirements Independent of physical spindle size Performance of disks (even unintelligent arrays) accelerated by network-wide caching Offloads much of disk read-access Makes existing disks suitable for more applications Higher application throughput and SAN performance Storage-related tasks offloaded from application hosts More responsive applications More responsive production LANs Reduction in planned downtime means More time available for online processing More time available revenue generating workloads Improved staff productivity/coverage Achieve 25X greater coverage through consolidation of distributed storage assets Manage more with current staff Fewer people required to manage storage Centralized management Intuitive drag-and-drop GUI Selective privileges Enhanced security Only explicitly assigned virtual volumes are visible to hosts Rogue hosts are prevented from accessing or discovering neighbor�s resources

8. Performance Benchmark Results 104,000 I/Os per second Many, frequent transactions Small block size, 512 bytes Example: database applications 560 MB/sec Throughput Rate Few, continuous streams of data Large block size, 1MB Example: streaming video 208,000 I/O�s per second and 1.12 GB per second in minimum Softek configuration

9. Competitive Comparisons Softek Can Improve the Performance of Leading Hardware Based Competitors by 2x or More Softek Virtualization Can Extend the Life of Existing Hardware by Sitting Between It and the SAN Softek Virtualization Can Provide the Magnitude Greater Performance for Less on Alternative Hardware

10. Softek Delivers An Industry Leading 1TB SAN For $150K

11. Benefits of True Virtualization Improves ROI and lowers TCO Eliminates underlying complexities Expands storage without disrupting applications Eliminates hardware dependence Offers scalability and high availability at the least cost Centralized Disaster Recovery in one tool Extends life of existing assets Delivers the highest performance What is virtualisation. One type of storage. SAN is now two zones. SDS virtualises any type of storage IO passes from server, to switch, to SDS, checks cash. If cash hit sends it back through switch to server. If cash miss sends it to switch to storage. Return is storage to witch, sds, switch, server. Disk is 4miliseconds. Cash miss is 1.25miliseconds. Disc is slowest component. Improves ROI from networked storage pools Improved storage space utilization Double disk utilization by tapping into misallocated or unreachable capacity Higher application throughput and SAN performance Improved staff productivity/coverage Achieve 25X greater coverage through consolidation of distributed storage assets Free portions of premium-priced assets Variable cost-basis for production assets Reduces storage-related TCO Defer new storage and server acquisitions Reduced storage administration Manage more with current staff Less expensive hardware Easier to manage Protect investment of existing storage assets Retro-active investment protection Protect your initial investment Protection going forward Improve your bargaining power Eliminates underlying complexities No LUN masking; only two zones Expands storage without disrupting applications Dynamically provision storage among needy hosts on demand Eliminate downtime attributed to adding, replicating, reassigning, and upgrading storage Eliminate large % of planned server downtime Add, upgrade, reassign and replicate disks online Logical volumes behave just like directly-attached disks Facilitate non-disruptive technology insertion Negotiating leverage Better products Different standards New suppliers Eliminates hardware dependence Protects and optimizes existing storage investments Eliminates vendor lock-in and costly upgrades and expansions Avoid costs of major proprietary storage upgrades and expansions Works with heterogeneous platforms and hardware SAN implementations are trouble-free and less costly Uses resources effectively Support any open system OS Use mixed storage assets Leverage best-of-breed technology every time Scalability and high availability Eliminate business disruptions Unlimited expansion of channels, cache, and storage Storage control at the core of the network Scale servers as processing power is needed Scale the amount and class of storage Grow network infrastructure as needed Remove single points of failure Instant fail over Use redundant hardware Mirror to dissimilar storage devices Synchronous mirroring up to 10K away Delivers the highest performance There is no performance trade-off Scale performance as needed Performance leverages industry advancements Increase server and storage throughput Accelerate performance using cached storage network Maintain or exceed existing service levels What is virtualisation. One type of storage. SAN is now two zones. SDS virtualises any type of storage IO passes from server, to switch, to SDS, checks cash. If cash hit sends it back through switch to server. If cash miss sends it to switch to storage. Return is storage to witch, sds, switch, server. Disk is 4miliseconds. Cash miss is 1.25miliseconds. Disc is slowest component. Improves ROI from networked storage pools Improved storage space utilization Double disk utilization by tapping into misallocated or unreachable capacity Higher application throughput and SAN performance Improved staff productivity/coverage Achieve 25X greater coverage through consolidation of distributed storage assets Free portions of premium-priced assets Variable cost-basis for production assets Reduces storage-related TCO Defer new storage and server acquisitions Reduced storage administration Manage more with current staff Less expensive hardware Easier to manage Protect investment of existing storage assets Retro-active investment protection Protect your initial investment Protection going forward Improve your bargaining power Eliminates underlying complexities No LUN masking; only two zones Expands storage without disrupting applications Dynamically provision storage among needy hosts on demand Eliminate downtime attributed to adding, replicating, reassigning, and upgrading storage Eliminate large % of planned server downtime Add, upgrade, reassign and replicate disks online Logical volumes behave just like directly-attached disks Facilitate non-disruptive technology insertion Negotiating leverage Better products Different standards New suppliers Eliminates hardware dependence Protects and optimizes existing storage investments Eliminates vendor lock-in and costly upgrades and expansions Avoid costs of major proprietary storage upgrades and expansions Works with heterogeneous platforms and hardware SAN implementations are trouble-free and less costly Uses resources effectively Support any open system OS Use mixed storage assets Leverage best-of-breed technology every time Scalability and high availability Eliminate business disruptions Unlimited expansion of channels, cache, and storage Storage control at the core of the network Scale servers as processing power is needed Scale the amount and class of storage Grow network infrastructure as needed Remove single points of failure Instant fail over Use redundant hardware Mirror to dissimilar storage devices Synchronous mirroring up to 10K away Delivers the highest performance There is no performance trade-off Scale performance as needed Performance leverages industry advancements Increase server and storage throughput Accelerate performance using cached storage network Maintain or exceed existing service levels

12. Performance Benchmark Summary Performance on Scale or Better than Competition Highly scalable Priced at a Fraction of the Cost

13. Fears, Uncertainties, and Doubts (FUDs) �You are in the data path� EMC, IBM, HDS and every storage controller of merit recognizes the value of being in the data path Provide multiple ways (N+1 architecture, Snapshot, Alternative Pathing, Replication/Migration, AIM) to ensure business continuance and disaster recovery No single point of failure Multiple white papers and technical briefs are available Improve I/O and application throughput performance through advanced caching algorithms Unprecedented performance with 208,000 IOPS and 1.12 GB/sec throughput rates for our minimum SAN configuration Softek Virtualization can boost the performance of any storage subsystem in the network storage pool

14. �What is the real-life performance?� Real-life performance depends on the customer�s application and environment As with any testing, we maximized our system to attain the highest performance numbers possible Our recently completed benchmark disproves any theory that an in-band virtualization solution negatively impacts performance, in fact, it can boost performance to unprecedented levels for systems such as EMC, HDS Our competency center is designed for customer demonstrations, testing, and benchmarking Fujitsu Softek encourages any customer to bring in their own application servers, storage subsystems, and business applications and test the performance in our center under real-life conditions Softek Virtualization�s performance will continually increase as technology advancements are made in the network hardware such as HBAs, processors, etc. Instead of replacing large, expensive storage subsystems like EMC or HDS (costs range from $250K to over $1M) in order to take advantage of the latest technology and performance, Softek Virtualization only requires a minimal asset investment (replacing the HBAs will cost $7K) Fears, Uncertainties, and Doubts (FUDs)

15. �Why a QLogic Card?� QLogic makes a significant contribution to the performance benchmark As QLogic (or another company) releases the next generation of HBAs and switches, Softek Virtualization will immediately be able to take advantage of the performance increases for minimal cost Fears, Uncertainties, and Doubts (FUDs)

16. Fears, Uncertainties, and Doubts (FUDs) �You use NT servers for storage management� Actually, Softek Virtualization uses NT for the management console GUI interface platform The technology to manage zoning, I/O, application throughput, mirroring, etc. is a custom application NT is a proven commercial operating system, with advances being made everyday First to market with all the latest, best and relatively inexpensive hardware No vendor lock-in to a proprietary platform Price/performance continuously advances Wealth of qualified I/O, networking interfaces, and 3rd party software Like all networks, customers must build in redundancy Implement data, path, and platform replication through independent storage control nodes running asynchronous to each other with N+1 redundancy

17. �Why an Intel Chip?� Intel is a great chip in the market Many of the world-class storage providers, including Hitachi�s Lightning are running Intel chips in both the front-end (channel adapters) and back-end (device adapters) Softek Virtualization will immediately be able to take advantage of the new advances in Intel technology and performance for a minimal cost Fears, Uncertainties, and Doubts (FUDs)

18. Softek SANView SANView is designed for customers who need to visualize, monitor and manage Storage Area Network assets SANView is a SAN management product that provides: a topology view of the SAN the most information for proactively monitoring and managing SAN performance, communication, and connectivity.

19. Today�s SAN is Difficult To Manage

20. Real-time & Historical Performance Statistics SANS were too complex and too expensive to manage � now can be managed from a centralised tool. To prevent bottlenecks and help you plan SAN management, SANView gives you a view of the I/o through individual ports and aggregates it for the entire switch. Not only do you get point in time information, but SANView tracks the historical performance. This can give you insight into when/where peak loads are occurring and allow you to rezone as needed. SANS were too complex and too expensive to manage � now can be managed from a centralised tool. To prevent bottlenecks and help you plan SAN management, SANView gives you a view of the I/o through individual ports and aggregates it for the entire switch. Not only do you get point in time information, but SANView tracks the historical performance. This can give you insight into when/where peak loads are occurring and allow you to rezone as needed.

21. Rapid Problem Resolution Fix the problemFix the problem

22. With Fujitsu Softek, the possibilities are truly infinite.With Fujitsu Softek, the possibilities are truly infinite.