2Q2008 System z High Availability – Parallel Sysplex TGVL: System z Foundation 1 System z High Availability – Value of Parallel Sysplex IBM System z z10 ECz10 BC David E Brown
2Q2008 System z High Availability – Parallel Sysplex TGVL: System z Foundation 2 High Availability as a Concept Parallel Sysplex Architecture Value Of Parallel Sysplex Data Sharing Learning Objectives
2Q2008 System z High Availability – Parallel Sysplex TGVL: System z Foundation 3 High Availability as a Concept 1.High Availability as seen by the end user – Application Driven Bank Teller Stock Broker Internet user 2.High Availability Metrics AvailabilityDowntime 90%36.5 Days 99%3.6 Days 99.9%8.7 Hours 99.99%52.5 Minutes %5.3 Minutes
2Q2008 System z High Availability – Parallel Sysplex TGVL: System z Foundation 4 High Availability Challenges Unplanned Outages Hardware Failure – Servers, Storage, Network Software Failure – Operating System, Database, etc Customer Application Failure Datacenter Disaster Planned Outages Hardware Upgrades Software Upgrades Data Center Moves
2Q2008 System z High Availability – Parallel Sysplex TGVL: System z Foundation 5 Server Clustering -The Challenge
2Q2008 System z High Availability – Parallel Sysplex TGVL: System z Foundation 6 If you were the Architect - Server Clustering? 1.All servers can access (read/write) Databases (Serialization of Updates – Data Sharing) 2.All servers can be managed from one console 3.Scalable to very large Capacity 4.Applications can run on all servers 5.Some Servers can be taken out of service (either planned or unplanned) without affecting application availability 6.Servers can run different levels of Operating System and Middleware Some thoughts
2Q2008 System z High Availability – Parallel Sysplex TGVL: System z Foundation 7 Data Sharing – DB/2 Buffer Pool Management Log File DB2 Log
2Q2008 System z High Availability – Parallel Sysplex TGVL: System z Foundation 8 Data Sharing - DB/2 Buffer Pool Management DB2 Log
2Q2008 System z High Availability – Parallel Sysplex TGVL: System z Foundation 9 Parallel Sysplex Hardware –Time Synchronization –I/O Connectivity –Coupling Facility Software –XCF/XES –WLM CF Operating - System μ- code (Microcode) –CFCC (Coupling Facility Control Code) NOTE: z/OS and Middleware may be level n, n-1, or n-2. WLM manages workloads across all members of a Parallel Sysplex according to the customers “Business” priorities.
2Q2008 System z High Availability – Parallel Sysplex TGVL: System z Foundation 10 Poll Question CFCC runs under z/OS in a CF. (Yes/No)
2Q2008 System z High Availability – Parallel Sysplex TGVL: System z Foundation 11 Parallel Sysplex Physical View Logoical View Simple! self managed
2Q2008 System z High Availability – Parallel Sysplex TGVL: System z Foundation 12 The Link Technology … Inter System Channel (ISC) –Original Coupling Channel –Provides long distance connections Internal Coupling (IC) Channel –Fastest coupling performance –Reduced complexity –Increased reliability Integrated Cluster Bus (ICB) –High bandwidth link for short distance –Fastest interconnection link –Improved processor utilization PSIFB (Parallel Sysplex Infiniband –Open standards –Very high bandwidth –6 GB/Sec –Longer distance compared with ICB (150M)
2Q2008 System z High Availability – Parallel Sysplex TGVL: System z Foundation 13 Capacity beyond largest CEC Parallel Sysplex Summary
2Q2008 System z High Availability – Parallel Sysplex TGVL: System z Foundation 14 Learning Points – System z High Availability 1.Parallel Sysplex is a High Availability Solution implemented via hardware, software, and microcode 2.A Coupling Facility (CF) provides Serialization and buffering Services to members of a Parallel Sysplex – CFCC is the Operating System for all CFs 3.Sysplex Timers (and/or FTP) assures synchronized clocks for all members of a Parallel Sysplex – Required for Data Recovery 4.CF to Server connections can be ISC, ICB, PSIFB, or IC links – IC links are virtual links when CF is LPAR on the same server 5.Parallel Sysplex protects end users from Planned or Unplanned outages 6.WLM manages workloads across all members of a Parallel Sysplex according to the customers “Business” priorities. 7.z/OS and Middleware may be level n, n-1, or n-2.
2Q2008 System z High Availability – Parallel Sysplex TGVL: System z Foundation 15 Break Time