A lot of time has passed since that was true. We are using cache on RB1000 with CF drives, it works great. Modern CF drives don't have this problem that you are worried about.....or a miniIDE connector for a 2.5' HDD for cache...I wonder who will keep cache on CF or microdrive...this is stupid.
http://www.routerboard.com/pdf/tests_jan_2008pps.pdfI think you should describe how you tested throughput and make comparision with other RBxxx.
And you should use RouterOS BWtest too.
I have 50Gb of cache. Do you have 50GB CF?A lot of time has passed since that was true. We are using cache on RB1000 with CF drives, it works great. Modern CF drives don't have this problem that you are worried about.....or a miniIDE connector for a 2.5' HDD for cache...I wonder who will keep cache on CF or microdrive...this is stupid.
yes, there are 64Gb:
http://gadgets.fosfor.se/64-gb-compact- ... m-samsung/
I can tell you that my recent experience in using even brand name CF card as cache storage has not been pretty. Are you implying that today's CF card are capable of 6 billion write cycle ? If so, can you please recommend a brand ?
A lot of time has passed since that was true. We are using cache on RB1000 with CF drives, it works great. Modern CF drives don't have this problem that you are worried about.
Last time I checked it seemed like 50 percent of our traffic was http on my main router. I have Mikrotik redirect/DST-NAT port 80 traffic to a Squid. I typically see 20-30 percent byte hit or bandwidth savings on that traffic. It seems to speed up some slow websites as well. On another site with a 6mbps link(we are in process of updating) at peak times we can see uplink maxed at 6mbps but out interface occasionally jumps and hangs at over 7mbps. CPU load on the Squid box is usually 1-3 percent. It has a single SATA drive.The way proxy works is it caches the static object that your customer downloads, to serve it later when another client requests the same exact file. this can only work for static non-changing objects, which are images and static html files (which are becoming a rarity).
1. If your customers browse all the same sites - there are not many writes, just reads.
2. If your customers browse different sites all the time - there is no use for cache.
Even if you have a lot of data in your cache - it doesn't mean it's used all the time.
Squid Object Cache: Version 2.6.STABLE17
Connection information for squid:
Number of clients accessing cache: 600
Number of HTTP requests received: 89107264
Number of ICP messages received: 0
Number of ICP messages sent: 0
Number of queued ICP replies: 0
Request failure ratio: 0.00
Average HTTP requests per minute since start: 1735.3
Average ICP messages per minute since start: 0.0
Select loop called: 1155379705 times, 2.667 ms avg
Cache information for squid:
Request Hit Ratios: 5min: 43.1%, 60min: 43.4%
Byte Hit Ratios: 5min: 25.0%, 60min: 15.8%
Request Memory Hit Ratios: 5min: 0.7%, 60min: 1.0%
Request Disk Hit Ratios: 5min: 37.0%, 60min: 36.3%
Storage Swap size: 88501112 KB
Storage Mem size: 7876 KB
Mean Object Size: 21.79 KB
Requests given to unlinkd: 0
Median Service Times (seconds) 5 min 60 min:
HTTP Requests (All): 0.05633 0.05633
Cache Misses: 0.19742 0.15888
Cache Hits: 0.01235 0.01035
Near Hits: 0.16775 0.10281
Not-Modified Replies: 0.00091 0.00179
DNS Lookups: 0.02336 0.02683
ICP Queries: 0.00000 0.00000
I think esn.toolz was asking for a reduced price version Normis. You stated that the L6 license and cpu were a major part of the cost. Many people don't need your level 6 license but only need the features of the L4.it will come with a LEVEL6. it is written in the original postCan anybody tell me if the RB1000 gonna come with L4 licence?.