Specifications

B
Proxy Server: Squid
requests per second = 1000 / seek time
Squid enables more disks to be used simultaneously, increasing the number
of requests per second. For instance, if you have three disks with the same
seek time of 12 milliseconds, using the following formula will result in:
requests per second = 1000 / (seek time / number of disks)
= 000 / (12/3)
= 250 requests per second
In comparison to using IDE or SCSI disks, SCSI is preferable. Newer IDE
disks, however, have similar seek times as SCSI and, together with DMA-
compatible IDE controllers, increase the speed of data transfer without con-
siderably increasing the system load.
Size of the Disk Cache
It depends on a few factors. In a small cache, the probability of a HIT (find-
ing the requested object already located there) will be small, because the
cache is easily filled up so the less requested objects will be replaced by
newer ones. On the other hand, if 1 GB is available for the cache and the
users only surf 10 MB a day, it will take more than 100 days to fill the cache.
Probably the easiest way to determine the needed cache size is to consider
the maximum transfer rate of our connection. With a 1 MB/s connection, the
maximum transfer rate will be 125 KB/s. If all this traffic ends up in the
cache, in one hour it will add up to 450 MB and, assuming that all this traf-
fic is generated in only 8 working hours, it will reach 3.6 GB in one day.
Because the connection was not used up to its maximum capacity (other-
wise we would have procured a faster one), we could assume that the total
amount of data going through the cache is about 2 GB. In the example, to
keep all the browsed data of one day in the cache, we will require 2 GB of
disk space for Squid.
Summing up, Squid tends to read and write smaller blocks from or to the
disk, making it more important how fast it detects these objects on the disk
than having a fast disk.
RAM
The amount of memory required by Squid directly correlates to the amount
of objects allocated in the cache. Squid also stores cache object references and
frequently requested objects in memory to speed up the retrieval of this data.
The memory is one million times faster than a hard disk. (Compare the seek
171SuSE Linux – Firewall on CD2