Aggregation File System design about java_server

[harryle95]

Consider the following requirements:

• Save data associated with Content-Server PUT
• Data can be retrieved by client using Station ID
• Data expires after 30s of inactivity or if not the last 20 most recent updates
• Data can be backedup and restored if needed

[harryle95]

Design:
Map<String, Map<String, Map<String, String>>> archive: file system storage (in memory)
Map[ContentServerID] = Map<FileName, Map<FieldName, String>

• If PUT request for the same file -> overwrite previous file.
• Fields:
  • Updated: whether an update to the main database using this data has completed succesfully
  • Deleted: boolean flag on whether the data should be removed
  • Content: body of a PUT request.

Map<String, String> database. Where the key is station ID and value is content

PriorityBlockingQueue taskQueue - task queue

LinkedList updateList -> list of updates

[harryle95]

ExecutorService connRevPool = Executors.newCachedThreadPool();

Connection receive threadpool:

• Retrieve request message -> Update TimeStamp
• Classify - GET/PUT
• Handle GET:
  • if empty -> Return empty body, do not submite to taskQueue
  • otherwise -> Submit to task Queue
• Handle PUT: submit to task Queue
• Get future from taskQueue submission.
• Poll the future and send the result back to client

[harryle95]

ThreadPoolExecutor executorPool = new .... // Only single thread executor. Use Priority queue

[harryle95]

ScheduledExecutionService schedPool ... //Executed the following:

• Run backup
• Delete outdated files