[Top] [All Lists]

Re: productivity?

2011-08-25 09:05:35

Sabahattin Gucukoglu wrote:

The answer has got to be a new SMTP extension, or pair of extensions, to do maximum throughput sending on one connection and multi-message transaction advisory mechanisms to support co-operation between bulk software and the MTA for knowing the delivery load in advance, thus neutralising the discussion over server administrative policy, port or connection slot exhaustion, client selfishness, time to wait until shutdown, event-driven programming vs process-based servers, OS wars and all the other nonsense. We need an improved checkpointing facility. That's what we should do, I think.

+1. Or we can begin with a basic tenet of "do no harm." It is obvious the issues of a well vetting system perform for a past era where hubs were more dominant and using ideas of connection sharing were isolated to well known connections, conflicts in a modern world where there is less hubbing, more peer to peer and anonymous connectivity. Just consider how Facebook has become a new high mail sender where it needs to improve its client side loads but its distribution is towards a larger pool of smaller smtp servers. If it was assumed that every facebook user had a common host account, i.e., then we back to the centralization and hub-hub concepts where this stuff was ideal.

So its an issue of recognition that using this across the board has a direct impact with most of the SMTP servers which there are significant amount of the smaller systems combined. And it may be just a matter of costing one to scale up or scale out by 1 or 2 CPU units. Small? Sure, but dollars are dollars and its harder to justify expense these days especially if it was discovered that the extra expense is due to CS clients.

"Boss, we can reduce the need to buy more computers or improve the ones we
      have simply by disconnecting these connection hogging clients."

Don't you think that will get the attention of most cost cautious managers?

Who knows? But I do know that alienation does not help anyone and perpetuates lack of cooperation as a group and continue conflicts with the SMTP specs.

PS: The C10K document is inspired, and I'm a Unix person. My stuff is event-driven. Just so you know. :-)

Better and in many aspects ideal, its still open to contention tuning issues :)


Hector Santos

<Prev in Thread] Current Thread [Next in Thread>