Optimize request forwarding for accelerator setups

This work aims at providing optimized request forwarding for accelerator setups by collapsing multiple requests for the same object into one backend server request.


Table of contents

  1. Documentation
  2. Progress
  3. Configuration
  4. Internal design
  5. Known issues and shortcomings

Documentation

In accelerator setups it is desireable if the number of connections to the backend web servers is minimized. However, Squid being designed primarily for forward proxy operation does not take this into consideration. As a result there may be storms of requests to the backendserver if a very frequently accessed object expires from the cache or a new very frequently accessed object is added.

To remedy this situation this patch adds a new tuning knob to squid.conf, making Squid delay further requests while a cache revalidation or cache miss is being resolved. This sacrifices general proxy latency in favor for accelerator performance and thus should not be enabled unless you are running an accelerator.

In addition an option to shortcut the cache revalidation of frequently accessed objects is added, making further requests immediately return as a cache hit while a cache revalidation is pending. This may temporarily give slightly stale information to the clients, but at the same time allows for optimal response time while a frequently accessed object is being revalidated. This too is an optimization only intended for accelerators, and only for accelerators where minimizing request latency is morer important than freshness.

Progress

2006-09-27 Bug #1780: collapsed_forwarding and Vary
Need to deal with Vary when collapsing multiple requests for the same URL.
2006-05-18 Merged into Squid-2.6
The collapsed forwarding support has been merged into the upcoming Squid-2.6 release.
2003-07-28
Final release for Squid-2.5. Fixes problems with stuck objects, ims related bug fixes, window based refreshes and more.
2003-06-21 First public release
First public release as a patch to Squid-2.5
2003-06-18 Final delivery
Final delivery of the changes sent to customer
2003-06-15 Initial delivery
Initial delivery of the changes sent to customer
2003-05-18 Project signed
The project was initiated

Configuration

collapsed_forwarding on/off (default off)
This option enables collaps of multiple requests for the same URI to be processed as one request. Normally disabled to avoid corner cases with hung requests, but there can be large benefit from enabling this in accelerator setups where the web servers are reliable.
refresh_stale_window interval (default 0)
This option decreases latency on collapsed forwarding by initiating a revalidation request some time before the object becomes stale. This avoid having more than one client wait for the revalidation to finish.

Internal design

This effect is acheived by the following primary changes

  1. On cahce misses the new object is made immediately public, allowing new requests to attach to the pendign request. If the object is later found to be private then the attached requests will all detach and initiate new requests.
  2. On cache revalidations entry->mem_obj->ims_entry is set to the StoreEntry of the IMS query, allowing additional requests to optionally attach to the same backend IMS query.
  3. To avoid objects which for one or another reason gets stuck and does not receive a reply in a timely fashion entry->mem_obj->refresh_timestamp is used to avoid objects which has been pending for IMS reply more than 30 seconds.
  4. The optional refresh_stale_window also uses the same refresh_timestamp field to keep track of if a revalidation has been initiated or not.

The stale_hit effect uses the same mechanism by disregarding that the object is stale if there is already a cache revalidation running (ims_entry set).

Known issues and shortcomings


Squid Now! Cache Now! Valid HTML 4.0! SourceForge
$Id: index.html,v 1.3 2006/09/27 11:25:49 hno Exp $