10 May 2012

SAP and Novell and IBM

 SAP Applications with the DB2 pureScale Feature on SUSE Linux Enterprise Server and IBM System x: A best-of-breed Solution Stack

Today’s rapidly changing business generates increasing demands for flexibility and continuous availability of business applications. IBM System x servers, SUSE Linux Enterprise Server optimized for SAP applications, the IBM DB2 pureScale Feature, and SAP applications provides are an excellent response to these demands.
SAP solutions on the DB2 pureScale Feature are horizontally scalable on both the SAP NetWeaver Application Server component and the DB2 database server tier. Optimal scaling is supported through specific technologies such as Remote Direct Memory Access (RDMA) capabilities over a high-speed cluster interconnect.

 The combination of IBM System x servers with SUSE Linux Enterprise Server for SAP applications is an excellent platform, offering the price and performance advantages of modern x86-based systems.

Figure: Overview of SAP applications on DB2 pureScale

 The horizontal scaling of the solution adds redundancy to the system, which on its own increases availability. Integrated DB2 cluster services take care of all servers and ensure continuous availability in the event of failures. During maintenance, you can seamlessly take individual components out of the cluster, ensuring that operations are not affected by the maintenance. The section DB2 pureScale Feature Use cases and Benefits for SAP Applications covers the availability, maintenance, and scalability aspects of this solution in more detail and explains the business value of these aspects based on use cases that are specific to SAP solutions. The performance characteristics are described in the section Performance considerations.

IBM, Novell, and SAP have closely worked together to tightly integrate the components. Specific optimizations have been added to the components to ease the initial setup and operations of the system. These optimizations and the resulting benefits are described in the section Integration and operation of the DB2 pureScale Feature with SAP Applications.
Main components of a DB2 pureScale instance

A DB2 pureScale instance consists of one or more DB2 servers, referred to as members. Each member runs a full DB2 engine with its own facilities, for example buffer pools, local lock lists, agent threads, and log files. The DB2 pureScale Feature is a shared data distributed database architecture, meaning each member has access to the same data for read and write operations. Synchronizing access to this data is required for database consistency, and being able to do so efficiently is essential for good database performance. The solution for ensuring database consistency while maintaining high levels of performance in the DB2 pureScale environment is based on DB2 for z/OS Parallel Sysplex, a shared-data DBMS that is known for its exceptional availability and scalability.

DB2 for z/OS Parallel Sysplex achieves its recognition largely because of its use of the coupling facility. The coupling facility provides a centralized location for managing locks and a global shared cache for data. Part of every DB2 pureScale instance is the pureScale cluster caching facility (CF) server, which provides the same centralized locking and global shared cache architecture. Whenever SQL processing on a member requires a lock, such as for updating a row on a page, it requests the lock from the CF. The CF determines whether the lock is available and if so, the lock is granted, and execution can continue on the member. Requests to the CF are also made when a member wants to read a page into its local buffer pool (LBP) or commit changes to a page. During a read request, if the CF has the page cached, it sends the page to the member, thus saving the member from having to read the page from disk. When the CF receives a page write request, the modified page is sent to the CF as part of the request and is cached in the group buffer pool (GBP). At this point, the CF has the most recent version of the page and proceeds to “invalidate” (mark as “stale”) any other copies of the page that exist in the LBPs of other members.

To eliminate the CF as a single point of failure, the DB2 pureScale instance is generally configured to use two CFs. In a two-CF configuration, all requests are sent to the CF that is designated as the primary CF and are also synchronously duplexed to the secondary CF. The secondary CF can then quickly take over as the primary CF if the primary CF fails. Communication to and from both CFs is optimized with extremely low latency communication performed over an RDMA-capable network fabric.

InfiniBand and 10 Gigabit Ethernet are supported for communication in a DB2 pureScale environment. InfiniBand and 10 Gigabit Ethernet support interrupt-free RDMA, which plays a critical role in the scalability of the DB2 pureScale Feature. With RDMA, each member can directly access memory on the CF, and the CF is able to directly access memory on each member, without using any CPU time at the server being remotely accessed. For example, during commit processing, all the pages that were modified during the transaction are sent to the CF. The CF then caches these pages and determines which (if any) of the other members have (now stale) copies of these pages in their LBPs. After obtaining this information, the CF uses RDMA to directly access the pages in the members’ LBPs and marks the pages as invalid. The ability to mark these pages as invalid with no interruption to the execution taking place on the member is critical for the scalability of write-intensive applications.

All servers in a DB2 pureScale instance are connected to the same disk storage through a storage area network (SAN). IBM General Parallel File System (GPFS) is used to provide access to the data with high scalability, high availability, and seamless capacity expansion. GPFS is shipped as part of the DB2 pureScale Feature and is installed and configured seamlessly during the DB2 pureScale Feature installation. The db2cluster command creates and manages GPFS file systems and automatically configures GPFS with known best practices for settings such as file system block size and file system cache size.

The DB2 pureScale Feature uses IBM Tivoli Systems Automation (TSA) and IBM Reliable Scalable Cluster Technology (RSCT) to automate failure detection and recovery. By offering a way to monitor the state of each of the components in a DB2 pureScale environment, TSA and RSCT help to provide continuous high availability. Upon failure detection, TSA and RSCT take the appropriate action to bring the failed component back to an operational state. Examples of actions include remounting file systems, restarting a DB2 member, and rebooting a server. Like GPFS, TSA and RSCT are shipped as part of the DB2 pureScale Feature and are installed and configured seamlessly during the DB2 pureScale Feature installation. You can easily manage TSA and RSCT after installation through the db2cluster command.

Figure 2 illustrates the topology of a DB2 pureScale instance with four DB2 members and two DB2 pureScale CF servers. One CF server is designated as primary (P) and the other as secondary (S).

Figure:. DB2 pureScale architecture

Integration and operation of the DB2 pureScale Feature with SAP applications

The DB2 pureScale Feature is tightly integrated into SAP applications. The integration covers all aspects of database usage with SAP applications: installation, simple database operation, system management, monitoring, and integration of HA for SAP central services.

The tight integration of the DB2 pureScale Feature into SAP applications allows you to perform a wide spectrum of database related tasks from SAP tools rather than with DB2 commands.

The DB2 database management system provides two very comprehensive scaling technologies: The DB2 pureScale Feature and the Database Partitioning Feature (DPF). DPF is a shared-nothing scale-out technology that has been supported for OLAP applications such as SAP NetWeaver Business Warehouse components for many years.

Figure . Managing initial setup and topology changes

SUSE Linux Enterprise Server as an optimized Operating System for SAP Solutions

Through collaboration with Novell, SAP selected SUSE Linux Enterprise Server as the only operating system for solutions such as SAP NetWeaver Business Warehouse Accelerator software, the SAP NetWeaver Enterprise Search application, the SAP Discovery System, and the SAP StreamWork™ application. Also the new SAP HANA™ application software is based on SUSE Linux Enterprise Server 11. SUSE Linux Enterprise Server is one of two reference platforms for SAP software development.
SUSE Linux Enterprise Server already includes the “sap-conf” and “sap-locale” packages, which prepare SUSE Linux Enterprise Server for the installation of SAP software.

SUSE Linux Enterprise Server for SAP Applications
Novell provides SUSE Linux Enterprise Server for SAP applications, which is optimized for mission-critical SAP software solutions. SUSE Linux Enterprise Server for SAP applications comprises software, maintenance, and support offerings.

SUSE Linux Enterprise Server for SAP applications version 11 service pack 1 differs from the regular SUSE Linux Enterprise Server (SLES) version 11 SP1 in that the SAP-specific version contains several additional product features, which are listed in Table

SUSE Linux Enterprise Server for SAP applications version 11 inherits all SAP hardware and software certifications from SUSE Linux Enterprise Server 11. These SAP certifications result from a quality assurance process of these products released for your mission-critical SAP applications. Hardware certifications (for example of IBM System x), and software certifications (for example of IBM DB2) are valid for the combination of the certified product and each of the Linux versions: SUSE Linux Enterprise Server and SUSE Linux Enterprise Server for SAP applications.

These SUSE Linux Enterprise Server versions support the latest multicore processor architectures from Intel and AMD, which are built in IBM System x and BladeCenter servers. SUSE Linux Enterprise Server for SAP applications is available only for the x86_64 architecture.

Novell and SAP have extended their support partnership with SUSE Linux Enterprise Priority Support for SAP applications. This unique Enterprise Linux support subscription integrates with SAP Solution Manager and entitles SAP users to receive Linux support through their established SAP support channel. Furthermore, this support offering leverages the SAP LinuxLab.

Figure: SAP support for DB2

 Application-transparent maintenance of a DB2 member or CF
The DB2 pureScale Feature helps minimize the outages for planned hardware or operating system maintenance.

To perform the maintenance on a host belonging to the DB2 pureScale cluster, first move the work off that host. For a DB2 member, use the DB2 QUIESCE command to force work off. Current work continues until the transaction ends and then transparently fails over to the configured failover member. To move the work to a different member other than the configured failover member, update the client affinity file before issuing the DB2 QUIESCE command, and trigger a configuration reload in the DBA Cockpit. For example, you might want to temporarily define a spare member as a failover target.

When there are no active connections on the member, the member is taken offline. Then you can take the member host out of the cluster if you want and perform maintenance. After themaintenance is complete, the host and the DB2 member can be reintegrated into the cluster. At the next transaction boundary, the SAP NetWeaver Application Server components automatically route work back to their primary member according to the client affinity configuration.

Figure. Maintenance of a member in a DB2 pureScale cluster

Maintenance on the CF works in a very similar way. First, stop the CF using the command db2stop . If you stop the secondary CF, the primary CF continues its work unaffected. If you stop the primary CF, the secondary CF takes over the primary role, transparent to the application, provided that the CF is in peer state. Depending on the planned maintenance, you can then put the host into maintenance mode. After you complete the maintenance, you can bring up the CF, which reintegrates transparently as the secondary CF.

Figure:. Maintenance of the CF


 Scale out or scale up? This is one of the most important questions that enterprise architects have had to answer since client/server architectures have been implemented in enterprise data centers. Scale out means providing more capacity by adding more servers, storage, and network components. Scale up usually means using bigger hardware with more CPUs, memory, network, and storage components. Both approaches have their advantages and disadvantages. Scale-out architectures work very well with modern x86 servers, in particular, with blade systems. The obvious disadvantage is that there are more physical servers to manage. Another challenge with scale-out architectures is concurrent access to data, in particular, concurrent write access. As a result, mixed two-tier environments are common, in which application servers are implemented on top of a scale-out infrastructure and the backend system, which is a database, is deployed on a large, scaled-up system.
With its scale-out architecture, the DB2 pureScale Feature provides redundancy and HA, including a complete HA solution for SAP applications. Except in a single-server solution, the failure of one component or member does not cause a service outage.

The DB2 pureScale Feature supports SUSE Linux Enterprise Server from Novell, a common platform for SAP software development. SAP trusts SUSE Linux Enterprise Server as an operating system for many SAP solutions, some of them exclusively available on SUSE Linux Enterprise Server. In addition, this Linux operating system leverages the latest x86 server technology from IBM. As described in this paper, you can implement small, midsize and very large SAP solutions with the DB2 pureScale Feature on the Linux operating system and x86 hardware and benefit from the price/performance advantage of that infrastructure.

The combination of SAP business solutions and the DB2 pureScale database product on SUSE Linux Enterprise Server is a best-of-breed solution that provides highest reliability and scalability at an very attractive price/performance advantage. Also, through the SAP Customer Support Network, IBM, Novell, and SAP jointly provide first-class enterprise support for the combined solution of SAP applications and the DB2 pureScale Feature on SUSE Linux Enterprise Server.

1 comment: