some thoughts on Microservice

Be in the Horzion project for months now, learn lots of new things and it might be a good time to document some learnings.

To switch from monolithic service to micro service means a total mightset change. The frist thing is to become a DevOps. You need to write/debug/test your code, build/deploy the code locally and remotely and monitor the service. This is very chanllenging in mature company because thses functions are isolated into different teams/organizations.

Here are some components in building microservice:

  • service register/discover
  • config management
  • (distributed) cache
  • log/alert/monitor
  • (distrobued) database
  • authentication for internal and external users

something need special handling:

  • how to handle exception: you can throws an exception in one service and hope another service catch it in old fasion.
  • how to pass data between services: if several services need to process the same request at different stage. if you need to merge the reply from couple services? or some service need the same intermedian data?
  • how to resolve service dependency: the business logic can be sync or async.
  • how to handle failover: if a service not available, will it retry and how?
  • how to document your service: this will impact how user can easily use the APIs

More to add later…

I love APIs

iloveapis-300x145

Here are some nice topics or “buzz words”.  The first one is from Amazon Micro-service dev-ops team. Actually, building micro service is a company culture not a practice. To achieve it, Amazon’s Jeff Bezos has a solution for this problem. He calls it the “two pizza rule”: Never have a meeting where two pizzas couldn’t feed the entire group. That is also true for a dev team. And also the team in charge of the whole life circle of the service: road map, technology, dev, QA, on-call, etc. In this way, the team is empowered and responsible for it.

Buzz words:

Another session is from Netflix, Daniel Jacobson. He describes how the micro servers overall picture in Netflix.

Buzz words:

Extra reading

Other buzz words:

keep-calm-and-love-apis-12

Java Queue implementations

In general, Queue is “A collection designed for holding elements prior to processing. Besides basic Collection operations, queues provide additional insertion, extraction, and inspection operations.”

There are two main catalogs:

Ref:

ConcurrentHashMap, Hashtable and Synchronized Map in Java

All these three class can be used in the multiple thread environment. Here is a summary of the difference:

  • Hashtable is since Java 1.0. The implementation lock the whole table.
  • Synchronized Map is since Java 1.2 under Collection framework. Please refer to the best practice. You also need to lock the whole map.
  • ConcurrentHashMap is since Java 1.5 and also a member of Collection framework. The map is partitioned into segments based on concurrencyLevel. So each segment can have a lock to improve the performance.

A detail comparison is at: http://javarevisited.blogspot.sg/2011/04/difference-between-concurrenthashmap.html

Hashtable

“This class implements a hash table, which maps keys to values. Any non-null object can be used as a key or as a value.”

Synchronized Map

“It is imperative that the user manually synchronize on the returned map when iterating over any of its collection views:”

Map m = Collections.synchronizedMap(new HashMap());
      ...
  Set s = m.keySet();  // Needn't be in synchronized block
      ...
  synchronized (m) {  // Synchronizing on m, not s!
      Iterator i = s.iterator(); // Must be in synchronized block
      while (i.hasNext())
          foo(i.next());
  }

ConcurrentHashMap

“A hash table supporting full concurrency of retrievals and high expected concurrency for updates. This class obeys the same functional specification as Hashtable, and includes versions of methods corresponding to each method of Hashtable. However, even though all operations are thread-safe, retrieval operations do notentail locking, and there is not any support for locking the entire table in a way that prevents all access. This class is fully interoperable with Hashtable in programs that rely on its thread safety but not on its synchronization details.”

Always use putIfAbsent instead of following code to achieve atomic action and better performance:

if (!map.containsKey(key))
   return map.put(key, value);
 else
   return map.get(key);

Ref