Accumulating Events

Background

Before describing how to accumulate events in AMI, it is important to understand the architecture of AMI, which consists of workers, local collectors, a global collector, a manager, and a client processes that run on different compute nodes.

The client assembles a graph which is pushed to the manager that then distributes it to the workers and collectors for execution. Nodes in the graph are "colored" worker, localCollector, and globalCollector. Each type of process is responsible for executing a subgraph of nodes of the appropriate color (ie workers execute nodes with the color worker, ...). Events in AMI are distributed to workers by Psana. AMI provides different types of graph nodes for accumulating events across this distributed architecture. Nodes which accumulate events are called global operations and GUI graph nodes are expanded into 3 underlying graph nodes.

PickN

PickN accumulates N events into a list. The GUI node is expanded into 3 underlying PickN graph nodes each with a color of worker, localCollector, and globalCollector. The underlying graph nodes have a different N.

Workers collect: N // num_workers

localCollectors collect: N // num_local_collectors

globalCollector collects: (N // num_workers) * num_workers

In the above example there are 4 workers and 2 localCollectors, if we wanted to collect N = ~25 events then each worker would collect 25 // 4 = 6 events, each localCollector would collect 12 "events" from the workers (ie 2 lists of length 6 each),  and the globalCollector would collect (25 // 4) * 4 = 24 "events" from the localCollectors (ie 2 lists of length 12 each).

The PickN will only update when N events have been collected. There is no guarantee on the order in which events are collected. 

RollingBuffer

RollingBuffer works similarly to the PickN, only it accumulates to a circular buffer, which shifts left after N elements have been accumulated. Intermediate results are returned.

For example assume N = 10 and we have 2 workers and 2 localCollectors. The workers and local collectors will both collect 5 events each and the global collector will collect 10 events. Once 10 events have been collected all elements in the list will shift left and the latest event will appear in the buffer at index 9.

worker1 buffer: [1, 2, 3, 4, 5]

worker2 buffer: [6, 7, 8, 9, 10]

localCollector1 buffer: [1, 2, 3, 4, 5]

localCollector2 buffer: [6, 7, 8, 9, 10]

globalCollector: [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]

worker1 receives NEW EVENT 11

worker1 buffer: [2, 3, 4, 5, 11]

worker2 buffer: [5, 6, 7, 8, 9, 10]

localCollector1 buffer: [2, 3, 4, 5, 11]

localCollector2 buffer: [5, 6, 7, 8, 9, 10]

globalCollector: [6, 7, 8, 9, 10, 2, 3, 4, 5, 11]

Accumulator

ReduceByKey