White Paper Maximize Storage Efficiency with NetApp Thin Provisioning and Symantec Thin Reclamation Jeremy LeBlanc; Adam Mendoza; Mike McNamara, NetApp Ashish Yajnik; Rishi Manocha, Symantec September 2010 WP-7111
TABLE OF CONTENTS 1 INTRODUCTION... 3 INITIAL PROVISIONING... 3 DAY-TO-DAY STORAGE USE... 4 TERMINATION... 5 2 THICK VERSUS THIN... 6 WHY THIN PROVISIONING... 6 3 MAXIMIZE STORAGE UTILIZATION WITH THIN PROVISIONING AND THIN RECLAMATION... 7 DOUBLE STORAGE UTILIZATION WITH NETAPP FLEXVOL... 7 GAIN ADDITIONAL STORAGE SPACE WITH SYMANTEC STORAGE FOUNDATION THIN RECLAMATION... 8 4 MITIGATE THIN-PROVISIONING RISK... 9 5 COMPOUND EFFICIENCIES... 10 6 CONCLUSION... 11 2
1 INTRODUCTION Over the last several years the unbridled proliferation of storage systems has come under increased scrutiny. Coupled with macroeconomic pressure, storage efficiency is a key initiative for almost every company. To address this initiative, companies are increasingly turning to new approaches to realize efficiency gains. In particular, thin provisioning is being rapidly adopted. The value is simple: allocate the capacity needed, then grow the space dynamically and on demand. Thin provisioning enables the application to be matched according to its actual storage consumption so the storage system can support more applications over a longer period of time, allowing storage administrators to delay storage purchases and reducing capital outlay. Applications and associated data go through a lifecycle of initial provisioning, day-to-day use, and termination. In some cases, applications continue in perpetuity, but most become obsolete, or associated projects reach their completion. Thin provisioning allows the application to grow as space is required. Thin reclamation allows the storage administrator to continuously reclaim and reuse space, avoiding or reducing additional capital expenditures. INITIAL PROVISIONING During the initial provisioning phase, applications are deployed, and associated storage is thin provisioned. Figure 1) Initial provisioning by administrator. 3
DAY-TO-DAY STORAGE USE During day-to-day use, aggregated storage pools enable each application to use resources as required. Thin provisioning allows this operation to be seamless. If an application outgrows its initial allocated space, additional space is borrowed from applications not using all of their initially allocated space, enabling dynamic growth and shrinking. Figure 2) Thin provisioning enables each application to use resources as required. 4
TERMINATION In the final phase of the application lifecycle, when a project is completed or an application is no longer used, the space is reclaimed and put back into the storage pool for use by other applications. Figure 3) Applications are deleted, and storage space is reclaimed. Thin provisioning alone provides huge gains in storage efficiency by optimizing storage utilization, helping storage administrators meet service-level agreements with the ability to quickly provision storage for new applications, and saving operational costs by upgrading storage in line with business usage. When combined with other efficiency technologies, such as deduplication, thin replication, and Symantec thin reclamation, the savings and efficiencies compound. This paper explores how NetApp storage efficiencies and Symantec Storage Foundation thin reclamation work together to provide maximum storage efficiency throughout the entire data lifecycle. 5
2 THICK VERSUS THIN With traditional thick storage provisioning, disk capacity is allocated to applications regardless of how much data is actually being written. Space requirements are rarely known early in a deployment, so, to make sure that their applications will have adequate capacity, users often present inflated estimates of capacity requirements, forcing administrators to overprovision the larger storage infrastructure. For example, a user requests a 2TB LUN for a new application that in reality only uses.5tb. Using thick provisioning, the 1.5TB allocated but not used by the application is wasted space because it cannot be used by any other application. This scenario, repeated over and over, leads to low storage utilization. From the storage administrator's perspective, it might appear that all of the storage on a system, which is 100% provisioned, has been given out, but at the application level, there might only be 25% utilization, creating 214TB of wasted space. Utilized Allocated But Not Consumed Free 100% 60% 40% 20% 0% 181 214 98 Thick Provisioned Allocated = 312TB Figure 4) Thick provisioning results in low storage utilization. WHY THIN PROVISIONING Thin provisioning separates the logical representation of storage from the underlying physical disks, making it possible to allocate more storage capacity to applications than is physically installed. With thin provisioning, the storage administrator takes an aggregate and creates volumes for different consumers and applications to use, but does not actually allocate any space upfront to those volumes. The space gets allocated only when the host application needs it. The unused aggregate space is available for the existing thinly provisioned volumes to expand or for use in creation of new volumes. Once the aggregate free space declines to a predefined threshold, the storage administrator can attempt to reclaim space from volumes that have obsolete data, can delete volumes that might be no longer in use, or can purchase more storage. By allowing as-needed provisioning and space reclamation, thin provisioning can result in better storage utilization and smaller capital expenditures on storage infrastructure. For example, instead of provisioning 312TB, only 146TB is provisioned. Of the 146TB, 98TB is utilized, and 48TB is allocated but unused. 249TB is returned to the storage pool for use as needed. 6
Utilized Allocated But Not Used Free 100% 249 60% 40% 20% 0% 48 98 Thin Provisioned Figure 5) Thin provisioning improves storage utilization. 3 MAXIMIZE STORAGE UTILIZATION WITH THIN PROVISIONING AND THIN RECLAMATION Thin-provisioning technology is synonymous with efficiency. It provides the storage administrator a way to substantially reduce overallocating storage to users and manage the capacity based on usage instead of allocated size. This allows the administrator to maintain a common, unallocated storage space that is readily available to other applications on an as-needed basis. But even with thin provisioning, there can be wasted storage if space from deleted blocks is not reclaimed. Combining thin-reclamation capabilities offered in Veritas Storage Foundation by Symantec with NetApp FlexVol thin provisioning allows storage administrators to compound the efficiencies of thin provisioning and truly maximize storage utilization. DOUBLE STORAGE UTILIZATION WITH NETAPP FLEXVOL The first step in the implementation of any storage system is the allocation of capacity to servers. NetApp FlexVol technology lets storage administrators create virtual volumes that can be managed and moved independently from physical storage, without disrupting ongoing operations. Unlike other thin-provisioning technologies, FlexVol is a storage virtualization technology that pools physical storage and enables storage administrators to dynamically grow and shrink volumes to meet exact capacity requirements. As volumes decrease, their capacity returns to the pool for allocation to other volumes as needed. In addition to improving storage utilization, FlexVol also helps reduce storage provisioning time from days to minutes and improves performance and reduces bottlenecks by distributing volumes across all available disk drives. 7
Data Stored in Dynamic Flexible Volumes 200GB 150GB 300GB 1TB 100GB 200GB 50GB Flexible Volumes (2TB Provisioned) RAID Groups Aggregate Physical Storage (1TB Total) Figure 6) NetApp FlexVol allocates storage on demand for LUNs and volumes across all protocols (FCP, iscsi, FCOE, and NAS) and operating systems according to application needs. As an example, the College of Saint Rose in Albany, New York, deployed a shared storage system based on NetApp products and technologies. Using FlexVol, its virtual servers are now provisioned in minutes, and LUNs are provisioned in seconds without disruption. Before, those actions took weeks and cost up to $20,000. Saint Rose can also provision its storage on the fly and is achieving utilization rates of to 85%: a significant improvement over the 20% it was getting before upgrading to the NetApp system. GAIN ADDITIONAL STORAGE SPACE WITH SYMANTEC STORAGE FOUNDATION THIN RECLAMATION Enabling physical storage to be allocated on demand as application writes are performed on thin LUNs provides huge efficiency gains. However, as the data lifecycle progresses and data is deleted from a host file system, the physical storage stays allocated to a thin LUN, resulting in increasing amounts of wasted storage. When Veritas Storage Foundation from Symantec is combined with NetApp storage, physical storage use is automatically "right sized" to the actual storage used by applications. Figure 7) Veritas Storage Foundation Thin Reclamation maximizes storage utilization by automatically reclaiming wasted space from thin LUNs, "right sizing" physical storage utilization to true consumption by applications.(graphic supplied by Symantec.) 8
The Veritas Storage Foundation Thin Reclamation API discovers storage that has been allocated because it has been written to in the past, but is no longer in use because the corresponding files have either been deleted or moved. Using a single command, storage administrators can trigger thin reclamation across individual LUNs, groups of LUNs, or across all LUNs in use by Storage Foundation within an array. Internally, the reclamation is carried out down to the granularity of the LUN page size, making it possible to reclaim even a very small piece of unused storage (as small as 8KB for some arrays). Utilized Allocated But Not Used Free 83 214 312TB Allocated 249 48 297 98 98 98 Thick Environment Thin Environment Thin with Symantec Figure 8) In a thin environment, waste is low to start with but grows over time. With Veritas Storage Foundation Thin Reclamation, storage use equals the actual application data. In this example, only the 98TB used by applications is allocated, freeing up the remaining 297TB of physical storage. 4 MITIGATE THIN-PROVISIONING RISK Reducing the perceived risks associated with thin-provisioning and thin-reclamation technologies and methods calls for a management solution that provides end-to-end discovery and visibility of storage resources, service paths, and related capacity analytics in context with applications, business units, data centers, and tiers. The management solution must help: Manage the risk of not having enough capacity to support all host allocations Determine how much capacity is enough Mitigate the impact of one or more hosts demanding their allocated capacity Continually optimize capacity based on true application usage The thin-provisioning promise is based on the fact that usage patterns (in large numbers) are predictable and that the administrator can slowly grow the physical size of the environment based on trending past usage. Including thin-reclamation factors enables administrators to more proactively and accurately plan their storage environment, improve storage utilization rates, and get the most value from their data center investments. Along with the additional and compelling efficiency gains from thin reclamation comes the introduction of additional risk into the environment, including: 1. How much spare physical capacity exists 2. How overprovisioned the environment is 3. How predictable management is in terms of capacity consumption 9
Most management systems are trying to help manage thin-provisioning risk by setting different "thresholds" on the different elements on the storage (to deal with factor 1) or limit the amount a volume can overprovision (to deal with factor 2). However, very few systems deal with the most unpredictable of all: factor 3. To deal with this, you have to analyze and detect the behavioral change of users to alert in advance that the assumptions on which you based your thin-provisioning environment strategy are no longer valid. SANscreen manages the risk of thin provisioning, taking all three factors into account. This is done by allowing the administrator to set thresholds in terms of "unpredictable" users instead of just capacity values. The system will then automatically and continuously monitor the thin-provisioned environment against these rules and alert when something is wrong. To further reduce the risk of running out of storage and prevent the misallocation of storage, a good management system must also continually optimize the underlying thin-provisioned environment. Veritas Storage Foundation thin reclamation enables the automatic reclamation of allocated but unused storage blocks, thereby reducing wasted storage costs. This level of control will raise the confidence level in the technology and enable more and more environments to take advantage of the huge efficiency gains introduced by combining thin-reclamation capabilities and thin provisioning. 5 COMPOUND EFFICIENCIES Together, NetApp thin provisioning and Symantec thin reclamation dramatically increase storage utilization. Combining these technologies with other NetApp storage efficiencies compounds the capacity savings and helps achieve maximum asset utilization. Table 1) Compounding capacity savings with NetApp storage efficiencies. Technology Potential Capacity Savings Deduplication Reclaim up to 95% capacity for full backups; 45% on average Thin Provisioning (FlexVol ) Liberate 20% to 33% of your capacity Double Parity RAID (RAID-DP ) Use up to 46% less capacity versus mirrored data Deduplication or RAID10 Reclaim up to 95% capacity for full backups; 45% on average Thin Provisioning (FlexVol ) Liberate 20% to 33% of your capacity 10 95% 95% 33% 95% 33% 95% 46% 33% 46% 46% 95% Deduplication Snapshot Copies Reclaim up to 95% capacity; capacity Deduplication for full backups; Reclaim up to 95% only capacity changed for full data backups; uses 45% on average 45% on average storage Deduplication Snapshot Copies Reclaim Snapshotup Thin Provisioning Copies to 95% capacity; capacity Thin Replication (FlexVol for Snapshot ) capacity; full backups; only changed data uses 45% copies capacity; only up changed to 95% only changed on average data uses 95% storage data uses storage Liberate 20% to 33% capacity with our diskto-disk data protection storage of your capacity Thin Provisioning Thin Replication Deduplication (FlexVol Thin Replication ) Snapshot Copies Double Reclaim Parity up to RAID 95% 95% Virtual Clones up to 95% capacity; (FlexClone ) Liberate (RAID-DP Thin up replication to 20% ) 95% to 33% up to 95% capacity capacity up with with to disk-to-disk our diskto-disk data protection capacity for full backups; only changed data using uses data protection of capacity your capacity with our diskto-disk data protection Use 45% up on to average 46% less writable storage virtual copies; only capacity versus mirrored changed data uses capacity data Double or RAID10 Parity RAID Virtual Clones (FlexClone Thin Provisioning Thin Replication ) Virtual Clones (FlexClone ) (RAID-DP (FlexVol RAID-DP ) ) up to using Use up to 46% less capacity up to versus 95% mirrored data or RAID 10 to using 95% Liberate Use up to 20% 46% to 33% less writable virtual copies; only capacity with our diskto-disk data protection of capacity writable your capacity virtual versus copies; mirrored only changed data uses capacity data changed or RAID10 data uses capacity Snapshot Copies Double Parity Virtual Clones (FlexClone ) FlexClone capacity; RAID up to capacity using writable virtual copies; only changed data (RAID-DP only (virtual changed clones) ) data uses uses capacity up to using Use storage up to 46% less writable virtual copies; only capacity versus mirrored changed data uses capacity data or RAID10 Thin Replication up to 95% capacity with our diskto-disk data protection Double Parity RAID Virtual Clones (FlexClone )
6 CONCLUSION Thin-reclamation and thin-provisioning technologies are synonymous with efficiency and provide significant savings on capital and operating expenses that lower storage TCO. When combined, they provide server and storage administrators ways to substantially reclaim storage space and reduce over-allocating storage to users, thereby managing capacity based on usage instead of allocated size. This allows administrators to only keep on the floor enough capacity as is required for short-term needs. The combined solution provides the ability to manage, reduce, and reclaim storage space and keep thin volumes thin over time, especially in situations where a large number of writes and deletes occur. This improves planning and reduces the overall IT footprint and management costs. Over time, these savings compound beyond the storage infrastructure to affect operational costs, organizational efficiency, and business productivity. Saving 50% on power, cooling, and space costs; reducing backup times from hours or days to just minutes; managing twice the data without increasing headcount; and cutting TCOs by 30% to 55% in Oracle, SAP, Microsoft Exchange, and virtual server environments can all be achieved with NetApp storage efficiencies and Symantec thin reclamation. NetApp provides no representations or warranties regarding the accuracy, reliability or serviceability of any information or recommendations provided in this publication, or with respect to any results that may be obtained by the use of the information or observance of any recommendations provided herein. The information in this document is distributed AS IS, and the use of this information or the implementation of any recommendations or techniques herein is a customer s responsibility and depends on the customer s ability to evaluate and integrate them into the customer s operational environment. This document and the information contained herein may be used solely in connection with the NetApp products discussed in this document. 11 Copyright 2010 NetApp, Inc. All rights reserved. No portions of this document may be reproduced without prior written consent of NetApp, Inc. NetApp, the NetApp logo, Go further, faster, FlexClone, FlexVol, RAID-DP, SANscreen, and Snapshot are trademarks or registered trademarks of NetApp, Inc. in the United States and/or other countries. Microsoft is a registered trademark of Microsoft Corporation. Oracle is a registered trademark of Oracle Corporation. Symantec and Veritas are trademarks of Symantec Boosting Storage Corporation. Efficiency SAP is with a registered NetApp Thin trademark Provisioning of SAP AG. and All Symantec other brands Thin or Reclamation products are trademarks or registered trademarks of their respective holders and should be treated as such. WP-7111-0910