All standard releases of SafeSquid are 32-bit applications built for Intel Architecture Hardware.
CPU and Physical Memory can impact the overall performance and scalability in large networks.
Using multiple NICs improves performance, but may not be critically important.
Large Hard Disk Storage space is required for storing SafeSquid's logs, reports and content cache.
RAID is not important, and can degrade performance in certain situations, if it impacts the Disk I/O speed.
Depending upon acceptable TCO limits, use CPU with multi-threading support. The system resources required for a SafeSquid based web filtering gateway, depend upon the overall features desired from such gateway. The RAM and storage resources impact the reliability and scalability, of such a gateway. The maximum number of concurrent users, content caching, duration of log retention, and reports, decide the quantum of utilization of the system resources. The CPU resources impact for process reliability and latency.
To reliably accomplish the various content filtering, caching and communication related activities, it must have enough Memory. As a thumb-rule it is ideally recommended to provide about 7 to 10 Mb of RAM per user for small networks. But for environments having more than 100 users, even 5 to 7 Mb per user should be sufficient. The tables below discuss this in some more detail.
Minimal Requirements for Features
|Base Product||Image Filter||URLBL||cProfiles||SawMill|
Concurrency based Resource Overheads
|Basic Client Servicing||Content Caching|
|Peak Concurrent Connections||RAM Overheads||Disk Storage for Logs per day||RAM overheads||Disk Storage|
SawMill Operational Overheads
|Users||RAM Overheads||Disk Storage Overheads|
Advisories for Standard Installations
SafeSquid® has a very low Total Cost of Owner-ship, and a very good ROI. In the long term most users prefer to extract more out of the fixed costs, by increasing the derived results. It is therefore recommended to use Hardware that can be scaled for RAM / CPU / NICs.
- Choose H/W that can scale for RAM / CPU, so that you may accommodate more users, over a period of time.
- Use Hard Disks with good seek/read/write speed, to reduce latency in case you plan to use large content disk-caches.
- If you expect a large traffic to be handled, it would be a good idea to use a GigaBit NIC. To increase security, or to cater to multiple networks it would be advisable to use 2 NICs or more.
- System Configurations that have easily accessible Hardware drivers for Linux are absolutely preferable, and would be useful, if you plan to increase redundancy by using Clusters.