Checkpoints for the
Deployment View
- Has expected throughput been examined and compared with expected capacity to anticipate
performance problems?
- If data is distributed across several nodes, how are updates coordinated? (This usually
involves some form of two-phase commit protocol). Is it possible for multiple copies of
the same object to become out of sync with one another.
- Is reliable transport of messages required? Secure transport of messages?
- Have the message patterns been examined to ensure that services which require more rapid
response are available locally (via a LAN rather than WAN)?
- In a system in which some servers are redundant (have backup servers), can more than one
server "think" it is primary? What happens then? How is the situation resolved?
Can no processors be primary at some point in time?
- Are the failure modes documented (for example, what happens if a server goes down?)
- Can faults in the network be isolated and diagnosed?
| |

|