| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Installation and Configuration
Installation isn't the hard part: Configuring more advanced features to make the cache do exactly what you want is where any real challenge will lie. Squid config is accomplished by editing the squid.conf filegreat for experienced admins and software hacks, but intimidating for true first-timers. Every commercial product we tested provides a CLI and/or GUI, essentially hiding any native config file under a friendlier user interface. Our subjective overall assessment of each is shown in the table rows labeled "CLI", "GUI", and "Target Audience". A browser-based GUI is the primary management interface for every product except InfoLibria (GUI now in alpha). Squid's CacheManager GUI is passive, not useful for making config changes. The NICS GUIs are clean, responsive, and in our opinion, the best for novice use. As an "Appliance", NICS exposes what Novell deems necessary and hides the rest. For this reason, we find CacheFlow and NetApp GUIs better suited for more experienced admins. But NetApp is at the opposite end of the spectrum: there is so much packed into this GUI that it becomes hard to navigate. The CacheFlow GUI is a nice hybrid that balances interface simplicity with the ability to perform complex management tasks. The following GUI snaps let you glimpse overall look-and-feel; see individual reviews for further examples.
Anyone planning to administer a cache remotely should be concerned about securing management traffic. We strongly recommend using SSH, available with InfoLibria, NetApp, and easily added to secure telnet or shell sessions used to administer Squid. SSL is also fine, but limiting encryption to login/password doesn't go far enough, in our view. In larger, more distributed networks, integrated management becomes a requirement. Every cache we tested, except Quantex, can be queried with SNMP and a third-party NMS. Kudos to InfoLibria for generating enterprise traps signaling cache-specific events. Others were limited to standard traps reflecting major system events like reboot and link failure. Squid and NetApp support SNMP MIBs that allow an NMS to really monitor cache operation. But even basic MIB-II counters can be used to graph traffic with a tool like MRTG. Compaq outshines the rest when it comes to systems management. The TaskSmart is shipped with both SNMP and WBEM-based management systems that enable detailed platform query, event reporting, and inventory management. Even if you don't use an NMS, you can still keep an eye on nearly every cache through email (and sometimes pager) major event notification, backed by system-level logs that identify what led up to the event. Only CacheFlow survived a week in our lab without at least one problem requiring reboot. We were pleased to see that all but one commercial product notified us of reboot by email and/or logged event. Only the Quantex gave us nothing to go on for problem diagnosis.
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||