Chapter 3. Topology and infrastructure 33
separate JVM, often across a wire). The data that is stored in the grid is spread across the
JVM instances where WebSphere eXtreme Scale containers execute.
Figure 3-1 Distributed cache topology
A distributed topology is appropriate for most mission-critical enterprise web applications
where managing heavy loads with large data, achieving linear caching performance, and
executing large and complex application code are required. This topology provides fault
tolerance and high availability features. The operation of the application and its server has no
impact on the grid containers. The separation of JVM instances can help to avoid contention
of system resources and ensure replication performance and system response time.
3.2.1 Catalog service placement
The catalog service controls partition and shard placement and routing for all clients, making
it a critical component of the grid. As such, the catalog service must be clustered for high
availability and must be taken into account during the planning phase of the grid topology.
Unlike the WebSphere Application Server deployment manager, ensure that at least one
catalog server is up and functioning at all times; your grid is likely to become inaccessible if all
catalog servers are down.
ANY A PPLIC ATION J VM
Ap plica ti on
Stand-alone or managed
WebSphere eXtreme Scale
Container
Stand-alone or managed
WebSphere eXtreme Scale
Container
Partition
ANY APPLICATION JVM
Ap pl icat ion
Catal og service
Cat alog
Server
Catalog
Server
Cat alog
server
Gr id cl ie nt Gr id cli en t
EIS
Caution: Be aware that problems might occur if you have catalog servers or container
servers running in WebSphere Application Server processes with the same name. For
example, hosting a catalog service in more than one node agent, or if you have containers
running in servers with the same name on separate nodes (that is, server1). For more
information, see 5.1.6, “Duplicate server names in WebSphere Application Server” on
page 83.