A3 implementation

[Brice Steven Nkengsa]

I am confused about three things:
- When are new requests generated?
- When do events occur(Arrivals & Departures)
- When are requests dequeued from queues (i.e. queue at CPU, GPU, etc..)

Can anyone clarify these?
Thanks

[William Cowan]

Post by Brice Steven Nkengsa
- When are new requests generated?
- When do events occur(Arrivals & Departures)
- When are requests dequeued from queues (i.e. queue at CPU, GPU, etc..)
Can anyone clarify these?
Thanks

I am a bit perplexed by this question. I thought that the generation of new events,
the acceptance of new events and what happens when requests are dequeued is handled
in a lot of detail in the assignment description.

There is one possible clarification I can think of.

1. Arrival events contain requests that are enqueued when the event is received.
2. When the requests reach the head of their queue they are dequeued.
3. When they are dequeued Departure events containing the enqueued request are put
into the event-set.
4. When the Departure events are received the requests they contain are enqueued at
the next server, or leave the system.

Bill

[Brice Steven Nkengsa]

Post by William Cowan

Post by Brice Steven Nkengsa
- When are new requests generated?
- When do events occur(Arrivals & Departures)
- When are requests dequeued from queues (i.e. queue at CPU, GPU, etc..)
Can anyone clarify these?
Thanks

I am a bit perplexed by this question. I thought that the generation of new events,
the acceptance of new events and what happens when requests are dequeued is handled
in a lot of detail in the assignment description.
There is one possible clarification I can think of.
1. Arrival events contain requests that are enqueued when the event is received.
2. When the requests reach the head of their queue they are dequeued.
3. When they are dequeued Departure events containing the enqueued request are put
into the event-set.
4. When the Departure events are received the requests they contain are enqueued at
the next server, or leave the system.
Bill

What I meant was when we start with two requests in the system when is
the next one generated (i.e. time). Is the interval delta_t between
generation of requests fixed?

[William Cowan]

Post by Brice Steven Nkengsa

Post by William Cowan

Post by Brice Steven Nkengsa
- When are new requests generated?
- When do events occur(Arrivals & Departures)
- When are requests dequeued from queues (i.e. queue at CPU, GPU, etc..)
Can anyone clarify these?
Thanks

I am a bit perplexed by this question. I thought that the generation of new events,
the acceptance of new events and what happens when requests are dequeued is handled
in a lot of detail in the assignment description.
There is one possible clarification I can think of.
1. Arrival events contain requests that are enqueued when the event is received.
2. When the requests reach the head of their queue they are dequeued.
3. When they are dequeued Departure events containing the enqueued request are put
into the event-set.
4. When the Departure events are received the requests they contain are enqueued at
the next server, or leave the system.
Bill

What I meant was when we start with two requests in the system when is
the next one generated (i.e. time). Is the interval delta_t between
generation of requests fixed?

A new event is put into the events set when
1. a request is enqueued in a server (Departure)
2. a cnxn request is enqueued in the user (Arrival, I think)
3. an Arrival event is extracted from the queue (Arrival)

The time at which new events will be extracted is determined by sampling either an
exponential distribution (Most arrival events) or a lognormal distribution (most
departure events). The parameters of the distributions, of which there are many, are
things that we vary in the experiment.

Hope that this helps,
Bill