With the advancements in DB2 server
clustering technologies (both Sysplex and pureScale) and the DB2 Connect
client drivers, it is now possible to achieve near continuous
application availability. Availability is now no longer about only
database server availability, but there is more focus on availability
of applications that access the data stored on the server. Having the
server up and running is not very meaningful if the applications
connected to the server suffer connection failures and downtime. And
customers rightfully have come to expect that their local and
distributed applications originating from around the world, stay up
and running 24X7.
DB2 Connect client driver availability
technology has grown leaps and bounds over the last few years –
this ensures that applications continue to function in case of server
unplanned outages as well as planned outages such as rolling upgrades
when members of a DB2 cluster are migrated one by one. Client driver
availability algorithms complement server availability technologies
to ensure highest availability in a distributed environment. A
combination of features around workload balancing and failover have
been built into DB2 Connect client drivers. While workload balancing
achieves higher throughput by distributing connections and
transactions to members with maximum capacity at any point in time,
failover reconnects applications to next best available member in
case of a connection failure.
In recent releases of DB2 Connect,
default values of properties that determine high availability
behavior have been changed to values that work well out of the box
for majority of customers. Given the number of products in the DB2
stack (application server, client driver, DB2) and myriad high
availability knobs at each layer, customers found it difficult to
achieve an optimal end-to-end configuration leading to outages that
could have been avoided in the fist place. Default value changes also
control resource consumption and allow faster recovery from failover.
For example, the default value of maxTransportObjets was changed from
-1 (unlimited connections) to a more reasonable value of 1000, that
prevents runaway applications from creating too many unnecessary
connections and deplete system resources.
Number of features were added for
easier problem determination in complex high availability setups. For
eg, new statistics were introduced to better manage socket
connections to clustered DB2. More details were added to the errors
that are returned to the application going against a cluster – that
provide the application more information on what kind of failover
happened and whether the application needs to take further action for
a successful transaction. For eg, reason codes and detailed message
texts were added to the -30108 error that notifies the application
that the transaction did not qualify for a seamless reroute to
another DB2 member and that the application should resubmit the
transaction for the reroute to occur.
High availability technologies built
into the DB2 server and DB2 Connect client drivers complement each
other and together have come a long way to satisfy customer's needs
for continuous availability of business critical data and
applications. To get the most availability for distributed
applications, we do recommend you to upgrade DB2 Connect client
drivers to the latest available levels.
