New Features in PSP2 for SANsymphony -V10 Software-defined Storage Platform and DataCore Virtual SAN Updated: May 19, 2015 Contents Introduction... 1 Cloud Integration... 1 OpenStack Support... 1 Expanded Set of Software-defined Storage Platform Services... 2 Deduplication and Compression for Space Reduction... 2 Veeam Backup Integration... 2 VDI Services... 3 Central Console for Multiple Server Groups... 4 Performance Management... 5 Additional Insights into System Performance and Health... 5 Flash Memory Optimizations... 6 Random Write Accelerator... 6 Summary... 7 Note: Some features are available as separately-priced options. Please consult with your DataCore-authorized solution provider for more details on how to put the new capabilities to work for you. 2015 DataCore Software Corporation. All rights reserved. DataCore, the DataCore logo and SANsymphony are trademarks or registered trademarks of DataCore Software Corporation. All other products, services and company names mentioned herein may be trademarks of their respective owners.
Introduction The PSP2 release of SANsymphony -V10 and DataCore Virtual SAN brings important enhancements described below. Cloud Integration OpenStack Support DataCore now brings its comprehensive set of Software-defined Storage services to private, hybrid and public clouds built with OpenStack. The massively scalable architecture offers organizations a convenient and standardized way to roll out and manage applications across large pools of compute, storage and networking resources, using a mix of hypervisors. OpenStack administrators and applications simply issue standard OpenStack Cinder commands to provision detachable block storage resources from DataCore nodes to the Nova compute nodes. The DataCore software responds by registering the Nova compute nodes as hosts, then serves virtual disks to them with the desired characteristics (size, redundancy, QoS). The virtual disks are connected to the compute nodes over iscsi or Fibre Channel, just like any host. Snapshots of these volumes are also controlled through OpenStack Cinder requests. In this way, large enterprises and cloud service providers centralize provisioning and management of existing storage resources using standard OpenStack tools, without incurring the expense to replace them with OpenStack-compatible equipment. Note: The DataCore Linux driver for OpenStack Cinder servers may be downloaded from the DataCore Customer Support site at: http://datacore.custhelp.com/app/answers/detail/a_id/1419. 1
Expanded Set of Software-defined Storage Platform Services Deduplication and Compression for Space Reduction In addition to the capacity savings from thin provisioning, virtual disks can now be deduplicated and compressed in the background to reduce the space they occupy on disk. The benefits of deduplication and compression are most evident with general file shares and backup volumes that contain many blocks of identical data; especially those virtual disks that are not very latency sensitive. Essentially, you configure pools specifically for virtual disks meant to be deduplicated, and set policies to schedule when the post-process deduplication should occur in the background. The storage space freed up through deduplication and compression defers the need to buy more storage. Considerations Please consult the SANsymphony-V Online Help for additional guidance on use of the deduplication and compression feature. http://www.datacore.com/ssv-webhelp/ Veeam Backup Integration DataCore works with popular 3rd party backup products to simplify end-to-end data protection and ensure rapid data recovery at a granular level. Products from companies such as CommVault, Microsoft, Veeam and others take advantage of DataCore online snapshots to centralize backups for a wide range of applications, hypervisors, operating systems and storage devices. The combined solutions offload backups from the hosts to minimize the impact on applications. 2
In PSP2, integration with Veeam Backup software relieves the burden of VMware vsphere snapshots on hosts. Consequently, backups can be taken more frequently to meet more stringent Recovery Point Objectives (RPOs) without compromising business productivity. Many of the steps typically necessary to coordinate snapshots for backups have been automated so they can be scheduled at regular intervals without manual intervention. Integration with backup / recovery tools complements DataCore s zero downtime, zero touch high availability storage solution, and helps organizations adhere to the 3-2-1 rule: Maintain at least 3 copies of your data Store them in 2 different formats Keep 1 copy off-site VDI Services PSP2 introduces wizards to automate the creation and management of stateful, highly available virtual desktops on top of Windows Server 2012 R2 with Hyper-V. This cost-effective, high-density VDI implementation is well suited for organizations seeking to deploy desktops at a departmental level (500 or below). It employs DataCore s high-performance caching, and thin provisioning to diminish the dependency on physical storage. In contrast to competing products, the solution uses fewer, lower-priced internal disks in place of more expensive, external storage, thereby significantly lowering the cost per virtual desktop. 3
From the new VDI Console, administrators first create a source image containing the base operating system and the desired applications. Then VDI clones are generated for individual persistent (stateful) virtual desktops. One can deploy the virtual desktops in a highly-available (HA) configuration with their data mirrored across two nodes, or contain them to a single node. In the event one of the servers in an HA pair goes down, Windows Failover Cluster automatically resumes the affected virtual desktops on the redundant node, minimizing the impact on users. Additional commands are available to delete VDI closes (virtual desktops) or reset them to the original state. Central Console for Multiple Server Groups Organizations with virtual storage pools distributed across multiple locations can now manage the different DataCore server groups from a single user interface. Primary data centers replicating to numerous remote sites or simply responsible for provisioning and monitoring these branch offices will enjoy this new capability. 4
In addition, separate levels of control can be assigned to administrators with different responsibilities across a hierarchy of server groups. For example, a master server group can provide storage to several departmental server groups, each with separate subordinate administrators. The new capabilities streamline control and delegate responsibilities over widely distributed storage infrastructures making it possible to centrally manage them with fewer people. Performance Management Additional Insights into System Performance and Health Additional visibility into the health and performance of the virtual and physical storage resources is provided in PSP2. Alerts from storage devices that support S.M.A.R.T. (Self-Monitoring, Analysis and Reporting Technology) are displayed in the System Health and Alerts screens to warn of potential problems before they become more serious. Essentially, they help predict failures due to mechanical wear and surface degradation, as well as notify administrators when components have become defective. The notices can be used to automatically trigger emails to personnel responsible for maintaining the storage infrastructure. Some of the escalated warnings related to space consumption and I/O latency provided for physical disks are now available for virtual disks. Different thresholds for attention, warning and critical thresholds may be set for individual virtual disks according to the importance of the applications that depend on them. The expanded toolset helps Isolate performance and disk problems before they become serious while reducing the time and expense to troubleshoot root causes of application slowdowns. 5
Flash Memory Optimizations New performance optimizations included in PSP2 result in faster response from PCIe flash memory and SSDs. The optimizations exploit the inherent quickness of these flash devices and their low-latency interconnects. The enhancements are most beneficial when the cache retrieves data from flash in a pool. Along with faster performance, they yield better utilization from these premium-priced solid state resources. Thus, fewer of them are needed to meet Service Level Agreements (SLAs). Random Write Accelerator The Random Write Accelerator option launched with PSP1 has been enhanced to allow enabling and disabling on active virtual disks served to hosts. Workloads characterized by many random writes, such as frequently updated databases, ERP and OLTP systems, benefit greatly from this option. The remarkable improvements in writes (see graphs below) are most pronounced when using spinning disks (~33X faster), but also result in up to 3.6x faster writes from solid state disks (SSDs). The optimizations effectively overcome undesirable delays incurred when updating widely dispersed data blocks across these storage media making low-cost spinning disks suitable for applications with intensive random-writes. Considerations The actual performance benefits you experience for any one application will vary depending on the percentage of random writes that make up that application s I/O profile and the types of storage devices you employ. There are also some tradeoffs discussed below. For these reasons, the optional feature can be enabled for individual virtual disks, allowing you to select when to apply the optimization. It s worth noting that Random Write Acceleration is not expected to improve the performance of workloads that are already largely sequential, although it will still work. Also note that optimizing the I/O pattern to the backend devices doesn t change the characteristics of the underlying storage. Any limitations in latency, throughput or IOPS will still apply. 6
Tradeoffs Random write acceleration is designed maximize performance and therefore uses more storage capacity than the amount of active data in the virtual disk. This extra space reduces the copying required and therefore yields greater performance by mitigating the penalty of random writes. Summary DataCore customers upgrading to SANsymphony-V10 PSP2 will enjoy significant enhancements that benefit not only conventional data center operations, but those IT organizations seeking the benefits of private, hybrid and public storage clouds. 7