Squid Proxy on SSD?

Hello guys!! this is my first time to make a thread here...

to the point, i just made a port from linux LUSCA Squid proxy to windows based for slow connection like mine... i want to ask if SSD would make any difference than HDD on reading cache files?

Squid Proxy i made would make mostly write based activity, as it is use to replace browsers cache to drives' cache..

would you recommend to make folder cache on SSD or just HDD?

sorry for the bad english

 

How big are you making this in MB's? Is it something that you can do with a RAM Disk instead? If there was any difference in HDD/SSD location, the RAM Disk should maximize that difference.

Also, if HDD/SSD based - can the 'cache' drive you use not be used as a boot/OS drive - make it handle just the cache (and in the case of the HDD - partition it to a very small size - relative to the total HDD capacity - as this will make it noticeable faster too.

Hope some of this helps.

 

hmm, i limited it at 10 GB... no i can't do it with ramdisk...

yes, it can be store in any partition drive.. i set it in D: partition on default though... i just wanted to know, would SSD make the cache being read considerably faster? without sacrificing performance of ssd itself?

 

As long as you have enough free space on the SSD (preferrably unallocated space) then yes, it will be faster than a HDD.

Partition not necessary in this instance (except to 'shrink' the single partition you have on the SSD to give the controller as much spare area as you possibly can, for sustained, high performance over time).

Just wanted to add that (guessing) this will be the most stressful workload you can put your SSD through (if I'm right that it is very 4K r/w orientated) - if this is true, then consider leaving a lot of unallocated free space so that the performance doesn't plummet on your SSD (30GB or more is not out of the question).

As I'm assuming that you now have the full capacity in use, this is what you should do (instead of a SE and partitioning right from the install of Win7) to prepare your SSD for maximum performance (with Win7 installed, of course):

Right click on Computer, click Manage, select Disk Management and right click on the C: drive.

Select Shrink and let it complete that process.

Right click on the unallocated partition we just made and select New Simple Volume.

Click next, next, next keeping all the defaults (and making sure that 'Quick Format' is selected) until we have a new partition made.

Leave the computer on, but idle (don't use it) overnight - this will let the TRIM command that the Quick Format initiated to run as fully as possible.

Make sure to disable any power savings features that put the computer to sleep, hibernate or otherwise shut itself off.

In the morning (or after a good 8-10hrs or more...) go back into Disk Management and simply delete this new partition we created.

Leave the unallocated capacity as unallocated to ensure the fastest SSD performance with the least WA possible for your setup.



Hope this helps.

 

I see. I'm using an M4 128GB and If I used 10 GB for cache ( Drive D ), I will have 50 GB left after shrinking 20 GB for unallocated space.. Would 20GB be enough for GC to kick in?

And would the M4 feel the reduced life expectancy significantly? Because of constant write?

Sorry for asking many question, I want to make sure I do the right thing and I'm very insterested on how Squid Proxy would perform on an ssd in windows

 

Edited my post #4.

If I'm right that the types of read/writes your cache will concentrate on is 4K r/w's, then yes, that is the most stressful way to use an MLC based SSD and will reduce life expectancy significantly and drastically lower the overall performance of the drive too. That is why I suggest as much unallocated space as you possibly can throw at it for this use.

See:
http://en.wikipedia.org/wiki/Write_amplification


Is there anyway for you to check/measure what disk access pattern your Squid Proxy will incur on the storage medium?

 

If I'm not mistaken, squid proxy on HDD is largely read / write on sequence, that's why if I have to throw my cache on an HDD, I would have use defragment every and now, to reduce access time for reading the file cache...

Unfortunately there's no way for me to make sure which disk pattern is being used on squid proxy, because it will be a pain for me to measure the file and access...

 

it is random and small in nature(think all those tiny little gif, js, css). But most browser would not have too many multiple outstanding requests going on in parallel(I think no more than 4).

I would not waste my money for this usage.