CHAPTER 9. The Enhanced Backup and Recovery Solution

Transcription

1 CHAPTER 9 The Enhanced Backup and Recovery Solution Based on the analysis performed on the existing backup and recovery environment in Chapter 8, this chapter covers the new backup and recovery solution. The Backup Assessment Using Quick Scripts (Backup Assessment) run in the previous chapter produced a tremendous amount of data about the existing backup and recovery environment, taking the guesswork out of the drawbacks to the existing environment and results in a more efficient new solution. (The tool used for the analysis is actually a spin-off of Backup Assessment but for simplicity purposes, I use Backup Assessment throughout this book.) Based on the statistics, the ideal backup and recovery solution can be crafted with predicted results. This chapter covers the new solution based on the Backup Assessment and some of the technologies integral to the solution, including de-duplication and a replicated disk library. The EMC disk library in this solution has a legacy with many names, including Virtual Tape Library (VTL), Enterprise Disk Library (EDL), DL3D (which you see in the upcoming screen shots), and many others. I use disk library, but you may see these names used in output documents from various tools and online when researching disk libraries. Although the recommended solution is based on the disk libraries, this chapter covers a backup and recovery solution ideally suited for virtualized environments, called Avamar Data Store. The section about the Avamarbased solution covers its capabilities related to source-based de-duplication, 111 From the Library of Gayle M. Noll

2 112 Chapter 9 The Enhanced Backup and Recovery Solution which greatly reduces network bandwidth required and has other advantages. New Disk Library Backup and Recovery Environment The new backup and recovery solution employs a lot of advanced technologies that replace the old and inefficient backup environment. The old environment did not handle the backup needs of the client and, in the case of some application servers, did not complete backup and recovery consistently. The old environment, for example, had many local tape drives connected directly to application servers. The review of the Backup Assessment output and the ROI analysis both made clear that the old solution, like many I see during the analysis process, was not getting the job done. The new solution employs a disk library that backs up applications like a tape library. This virtual tape functionality of the disk library does more than "fool" the backup software into thinking it is a tape, and not a disk, library. By emulating a lot of tape functionality, backup software does not have to be changed, so backup administrators can operate the disk library in the same way they operate a tape library. In addition, tape drives connected to the disk library can be written directly to tape. Many backup programs are supported by the disk library, so the operational change to employ the disk library is minimal while realizing the immense benefits of de-duplication and other technologies. Figure 9-1 shows the new backup environment.

3 New Disk Library Backup and Recovery Environment Application Servers and Backup Clients NYC Production Network Segment 35 Application Servers and Backup Clients Central Backup Server Storage Server WAN Fibre Switch Fibre Switch NJ Production Network Segment Storage Servers SAN Disk 2 LTO-4 Tape Drives with Encryption Replication WAN Disk Library w/ De-Duplication Fibre Switch Fibre Switch Disk Library w/ De-Duplication SAN Disk Figure 9-1 New Backup and Recovery Environment Figure 9-1 shows that the two locations, NYC and NJ, each have a disk library local to them. Each location has a series of clients across the top of the figure that back up locally to their disk library. In the old solution, there were roughly 50 local tape drives locally connected to the clients where local backup took place. The next level down in Figure 9-1 shows that each location has storage servers, also called storage nodes. The "Storage Nodes" shown in both NY and NJ receive data from all clients to be backed up and sent to the disk library. The data is then replicated from NYC to NJ and vice versa from disk

4 114 Chapter 9 The Enhanced Backup and Recovery Solution library to disk library. This bidirectional replication takes place after the backup is complete. The backup takes place to the local disk library during backup hours, typically beginning around 8:00PM, and is completed before business hours. When business hours begin, the bidirectional replication takes place between the disk libraries in NY and NJ. In addition, the NYC location has a central backup server, which is a NetWorker server, that contains metadata on all backups and recoveries that have taken place. It controls the backup process in both the NYC and NJ locations. This node contains the catalog database, which is metadata related to all backups. The NetWorker server contains the backup schedule and initiates the backup process. The 40 application servers shown in each location have a NetWorker backup client agent on them that communicates with the storage node to facilitate backup and recovery. The client communicates with the storage node continuously during this backup process by obtaining the data from the SAN and passing it on to the storage node, where it is then written to the disk library. The production data is stored on storage connected to the SAN. Numerous production servers run various applications in both locations. There is no change in the operation of the data storage for applications with the change in the backup solution. The tape library is shown in Figure 9-1 because tapes may need to be shipped to a secure location in some cases. The disk library data is replicated so the backup data is in two different locations. For this reason, the client does not need physical tape backup to ensure the data is shipped offsite. There are, however, some compliance requirements that may include tape shipment offsite. The tape library shown also includes encryption capability. In addition, the tape library may be required to read old tapes. Alternative New Backup and Recovery Environment Some advanced backup technology ideally suited to virtualization, but works with non-virtualized environments as well, could be part of the backup and recovery solution rather than only disk libraries covered in the previous section. The more advanced solution includes Avamar Data Store, which performs source-based de-duplication in a virtualized environment. The de-duplication takes place before data is transferred to the Avamar Data Store. In the original solution, the disk library performed target-based deduplication, which means the data was transferred to the disk library where

5 Alternative New Backup and Recovery Environment 115 de-duplication takes place. The following are overviews of target and source-based de-duplication: Target based de-duplication takes the backup stream from backup clients, moves the data through the network LAN and WAN, then through the fiber channel pathways to the target device, such as the disk library. Once on the target, the data is then reduced based on common data block chunks when the data is received in the data stream (in-band) or after the backup stream has been completely received at a later time (out-of-band). This solution allows for deduplication to be incorporated into existing backup environments with no retraining or modification of the backup process. Source-based de-duplication solves the problem differently in that the de-duplication process takes place before any data is transferred through the LAN or WAN infrastructure. Using software algorithms located on the client, the application monitors all new disk writes executed by the client and analyzes it for commonality of data chunks. Once the delta, or changed pieces of data, are identified, categorized, and globally compared with other prior client backups, only the unique, new chunks of data are transferred over the network. After initial seeding of the source-based data store, these deltas are typically reduced by more than 90%. This source-based de-duplication is essential in a virtualized environment. Physical resource issues are created by attempting to back up 20 or more virtual machines on a single ESX server with a limited number of networking ports. Backups may not complete within the backup window under these circumstances. In an example where there are four networking ports, each of the 20 virtual machines consider those 4 physical ports their own and use them for the backup. All the virtual machines are hitting those four ports simultaneously during a standard backup. By using source-based de-duplication, you can limit the data that will hit those four physical ports by more than 90%. This removes the limiting factor of throughput in the ESX server to achieve a successful backup within the backup window. Virtualization is not required for source-based de-duplication and nonvirtualized blades or servers can take advantage of this benefit in the form of greatly reduced LAN and WAN traffic. The disk libraries are not required if the Avamar Data Store units are used as the backup and restore solution; however, I included both solutions in the alternative solution to show that both solutions can co-exist.

6 116 Chapter 9 The Enhanced Backup and Recovery Solution The individual virtual machines run the Avamar client that enables them to perform source-based de-duplication backups. The disk library looks like a tape library to the backup software, but the Avamar Data Store uses a specialized backup interface that manages all aspects of the backup and restore processes. Figure 9-2 shows the alternative backup environment. 7 Application Servers and Backup Clients NYC Production Network Segment 30 VM s with Avamar clients residing on 3 ESX Servers Central Backup Server Fibre Switch Avamar Data Store w/ WAN Replication WAN Fibre Switch 5 Application Servers and Backup Clients NJ Production Network Segment 30 VM s with Avamar clients residing on 3 ESX Servers SAN Disk 2 LTO-4 Tape Drives with Encryption Replication WAN Storage Server Disk Library w/ De-Duplication Fibre Switch Avamar Data Store w/ WAN Replication Fibre Switch Disk Library w/ De-Duplication SAN Disk Figure 9-2 Avamar Data Store in the New Backup and Recovery Environment Figure 9-2 shows that the two locations, NYC and NJ. Each location has an Avamar Data Store and disk library local to them. Each location has a series of clients across the top of the figure that back up locally. In the old

7 Disk Library Operation 117 solution, there were roughly 50 local tape drives locally connected to the clients where local backup took place. The next section covers the operation of the disk library, which was the key to the first proposed backup solution. Disk Library Operation In this section, I configure the disk library (know by many other names and shown as DL3D in the figures, including creating a virtual tape library and virtual tape drives, and establish a connection to a backup server over Fibre Channel using worldwide names. I then initiate a backup from the backup server using a virtual tape drive and show the result of the backup and deduplication on the disk library. Figure 9-3 shows the home page of the Management Console of the disk library. Figure 9-3 Disk Library Management Console Home Page Some points of interest in Figure 9-3 are:

8 118 Chapter 9 The Enhanced Backup and Recovery Solution All backups to the disk library from clients were 608.3GB, and the disk library space consumed was GB. This represents a deduplication of 71.41%. At the end of this section, you see that these numbers slightly go up when I back up an additional client. Under Virtual Tape Libraries Data Services, de-duplication is enabled on three virtual tape libraries. I configure an additional virtual tape library shortly. Replication is Ready, but not enabled. NAS Data Services refers to disk library disk area that is used for backup to the disk library as a file system as opposed to a tape library. This is an alternative to writing to a virtual tape. Figure 9-4 shows the details of the three virtual tape libraries that currently exist on the disk library. Figure 9-4 Virtual Tape Library Configuration As Figure 9-4 shows, the disk library can emulate a maximum of 64 unique tape libraries. There is a maximum of 160 tape drives spread across the tape libraries. The disk library in this example is the smallest available from EMC at the time of this writing, so the virtual tape library and virtual tape capacities are the minimum numbers you ll see for any disk library. The

9 Disk Library Operation 119 three tape libraries have model numbers associated with them to emulate the devices shown in the figure. Figure 9-5 shows the process of adding a fourth tape library. Figure 9-5 Adding a Virtual Tape Library The virtual tape library being added in Figure 9-5 is an ADIC i500 with 30 slots to hold HP LTO-3 tapes. De-duplication is enabled on this virtual tape drive. Figure 9-6 shows creating virtual media, which are tape definitions.

10 120 Chapter 9 The Enhanced Backup and Recovery Solution Figure 9-6 Creating Virtual Media Tape Drives In Figure 9-6 I selected a tape media type of LTO-3 that will go in to the ADIC i500 tape library. I have now created the virtual tape library and the 20 tapes that will go into the library. The tapes have barcode range LTO001-LTO020. Figure 9-7 shows the process of connecting the backup server to the virtual tape library I just created.

11 Disk Library Operation 121 Figure 9-7 Connecting the Virtual Tape Library Under Add SAN Clients Group are the following key areas to fill in to connect the backup server to the virtual tape library: The Host is the backup server with the worldwide name shown in the figure. The Target is the disk library Node 1:FC3. The table of Device and SAN Clients shows a Type of VMC, which is the disk library, and Type VTD, which is the virtual tape drive. On the right of the figure are all the Target Ports available on the disk library. There are four Fibre Channel ports and six Ethernet connections on the disk library. Figure 9-8 shows the devices connected the Windows backup server.

12 122 Chapter 9 The Enhanced Backup and Recovery Solution Figure 9-8 Devices on Windows Backup Server Figure 9-8 shows several tape drives available for backup. The four HP LTO-3 tape drives are part of the VTL created earlier. Figure 9-9 shows initiating a NetWorker backup job.

13 Disk Library Operation 123 Figure 9-9 Initiating a NetWorker Backup Job The backup job shown in Figure 9-9 will back up the client group Default to the VTD I have specified. Figure 9-10 shows some of the specifics related to this backup job.

14 124 Chapter 9 The Enhanced Backup and Recovery Solution Figure 9-10 Backup Job Information The backup job uses an ADIC VTL and VTD \\Tape0. This virtual tape device looks like a physical tape device in Windows and the NetWorker program. Figure 9-11 shows the results of the backup on the disk library.

15 Virtual Disk Library and Backup Software Implementation 125 Figure 9-11 Backup Results on the Disk Library Figure 9-11 summarizes the backup results for all backups run to the disk library, including the small backup of group Default (shown in the earlier figures). The total data backed up before any de-duplication was GB but the de-duplicated space required to hold this data on the disk library was only GB. This represents a data reduction of 71.37%. The next section covers the implementation plan I follow when implementing a disk library for backup and recovery. Virtual Disk Library and Backup Software Implementation This section covers the planning and implementation of the disk library and backup software. In this case, EMC Networker is the software deployed; however, the disk library supports many other enterprise backup programs. The following three phases cover disk library-related work:

16 126 Chapter 9 The Enhanced Backup and Recovery Solution Phase 1: Disk Library Planning Collect detailed information about the amount of backup size needed on the disk library, backup window Service Level Agreement (SLA) requirement, and desired retention policy. Plan the desired tape library and drive configuration based on information collected. Plan and designate the required backup network segment and IP addresses on which the disk library will reside. Plan for call-home requirement configuration. Plan and designate the disk library Console Management Server Plan and design for off-site requirement via cloning or vaulting using the embedded storage node or media server functionality. Phase 2: Disk Library Implementation and Testing Configure the disk library environment based on the required tape library and drive requirement. Implement zones on all required backup servers and the disk libraries to see all configured virtual tape library and drives. Configure all needed SAN clients. Assign planned drives and robotics controller to the planned backup server. Assign planned drives to the designated media server or storage node. Verify the ability to see all assigned tape library and drives on designated servers. Load all required tape drives licenses depending on the backup software used. Integrate the disk library into the backup environment solution. Implement and Integrate the disk library design for offsite requirement.

17 Virtual Disk Library and Backup Software Implementation 127 Phase 3: Documentation and Perform Knowledge Transfer Produce a detailed material list, including serial numbers and support numbers. Compile a document with the steps performed to implement the disk library. Organize a review meeting to cover the completed implementation. The next three phases cover NetWorker-related implementation. Phase 4: NetWorker Planning Collect details about the backup size that needs to be protected, gather information about backup window SLA requirement, and desired retention policy. Plan the desired tape library and drive configuration based on information collected. Plan and designate the required backup network segment and IP addresses that will be used in the configuration. Plan, design, and designate the required platform on which the Networker server will run. Plan, design, and designate the required platform and the number of NetWorker storage nodes required to meet the backup window and SLA requirement. (Best practice at the time of this writing is to have one storage node for a 2TB data environment.) Plan, design, and designate the required platform and the number of dedicated NetWorker storage nodes required. (Best practice is to have a dedicated storage node once the data on a server is above 500GB.) Provide the minimum technical requirement for installation on the NetWorker server, the NetWorker storage node, the NetWorker dedicated storage node, and the NetWorker clients. Provide the minimum technical requirement for installation on Net- Worker application module. Verify that all required licenses are available before any installation and configuration is scheduled.

18 128 Chapter 9 The Enhanced Backup and Recovery Solution Plan and design for VMware backup environment either via VCB backup or VMware Guest backup. Plan, design, and designate the backup policy, schedule, and retention policy. Plan and design the offsite requirement. Phase 5: Implementation and Testing Configure the disk library environment based on the required tape library and drive requirement. Setup all required zones on backup server and the disk library to see all configured tape library and drives. Configure all needed SAN clients. Assign planned drives and robotics controller to the backup server. Assign planned drives to the planned media server or storage node. Verify the ability to see all assigned virtual and physical tape libraries and drives on designated servers. Load all required tape drive licenses depending on the backup software used. Install and configure NetWorker Server and apply all required patches. Install and configure NetWorker Storage Node Server and apply all required patches. Install and configure NetWorker Dedicated Storage Server and apply all required patches. Install and configure NetWorker clients and apply all required patches. Install and configure NetWorker Application modules and apply all required patches. Install and configure NetWorker VCB proxy host and apply all required patches for all VMware clients. Test backup and restore functionality for all installed and configured server and clients.

19 Summary 129 Implement and integrate the disk library into the backup environment solution. Implement and integrate the offsite design. Phase 6: Documentation and Knowledge Transfer Implement and integrate the offsite design. Produce a detailed material list, including serial numbers and support numbers. Compile a document with the steps performed to implement the disk library. Organize a review meeting to cover the completed implementation. Summary The advanced backup solution presented in this chapter, based on disk library technology, is much more operationally efficient and advanced than the old backup solution based on local tape drives. The next chapter shows that there is a compelling ROI associated with the new backup solution.

20 This page intentionally left blank