aiCache F.A.Q Page
- What kind of web sites benefit from aiCache the most?
- We're a busy site that is low on tech staff, can we get help with install and configuration?
- What kind of hardware do you recommend for an aiCache server?
- Why does aiCache only run on Linux and why 64bit ?
- Why aiCache when we have Apache and Microsoft IIS?
- We use memcached do we need aiCache?
- aiCache does work miracles with HTTP, what about HTTPS?
- What about analytics?
- Why aiCache when we have [insert a proxy web server here]?
- We personalize Web pages, can we still use aicache ?
What kind of web sites benefit from aiCache the most?
Short answer - all and any. News sites , community: message boards and blogs, social networking, rich media - on-line video, picture hosting , on-line shopping/merchant sites. This is by no means this is an exhaustive list, but the point is - all of the web sites will benefit from deploying of aiCache. You will spend less on HW, staff and hosting, will have faster site and happier visitors, better uptime statistics, be able to handle traffic spikes that used to take it down.
Main benefits are caching and off-loading of client request/response processing away from servers that were never designed to handle it...
We're a busy site that is low on tech staff, can we get help with install and configuration?
Absolutely, optional professional services engagement is available as part of purchase. We'd have our staff evaluate your sites setup to see how it can best benefit from aiCache, install and configure product in a turn-key fashion. Very little to no upkeep is required after installation, except regular chores of maintaining disk space etc.
What kind of hardware do you recommend for an aiCache server?
Any 64bit Intel or AMD processor-based system should do. Nowadays most of the 64bit processer are multi-core, meaning they have a number of separate CPUs on one piece of silicon. A quad core (a single quad core CPU or 2 dual cores) is where you get the most bang-for-your-buck - as aiCache is right-threaded. Only 4 threads are used to do most of processing (and 2 more are only used episodically), so that works most nicely with 4 CPUs.
Getting even more cores (CPUs) will not have nearly as much impact as spending additional money on extra memory. Sites differ in the amount of cacheable content, but you definitely want to make sure it fits into available RAM - so size the RAM accordingly. 8GB is a good start for most sites. As to what vendor to choose, we cannot recommend a particular one, for obvious reasons, but most companies have excellent luck with established vendors - such as HP, Dell, IBM and SUN.
Please note that aiCache is also available as preconfigured, ready to be started, pay-as-you go Cloud images, configured for most popular Cloud provider networks.
Why does aiCache only run on Linux and why 64bit requirement ?
Linux is only OS that offers the most efficient network IO model - multiplexed EPOLL mechanism. No other operating systems offer this particular mechanism and/or scale anywhere close to how EPOLL scales.
aiCache caches all of the cacheable responses in memory (RAM), never on disk (local or NFS etc) - so the more RAM you have in your server the merrier. 32bit systems are limited to offering around 3GB of RAM space per process (even though more than 4GB may be installed). 64bit systems do not have this limitation and there's no practical limit to just much memory might be available to a process. As of 2008 you can buy systems with up to 64GB RAM, in 2009 we expect systems with 256GB RAM to become available. Chances are you won't need anywhere close to this much RAM, but some larger sites do benefits from having this much capacity.
Why aiCache when we have Apache and Microsoft IIS?
aiCache doesn't replace the regular web servers - it only dramatically reduces the number of them that you need to support your web site. It also dramatically reduces the number of App and DB servers, file appliances etc.
aiCache offloads most of the load from your farm (we've seen sites with 98% offloading), so that most user requests are served right by aiCache, without ever having to touch your web servers, propagating in turn to your application servers, database servers and file servers.
Regular web servers are ill suited to supporting large volumes of requests or large number of users - most require a dedicated process per connected user, especially for dynamically generated content. This is the main reason the large web sites often grow out of control across the board - in total spend, number of servers, number of staff, amount of datacenter space, power and cooling that is required . And still, even with this many servers site have difficulties staying up.
We use memcached, do we need aiCache?
aiCache and Memcached are perfectly complimentary technologies. aiCache removes the majority of the requests to the web, app and database server BEFORE they reach the source environment.
Memcached decreases the time to generate original content from the source environment by front ending the database.Together they provide the optimal combination of dynamic content generation and distribution.
Memcached is a high-performance, distributed memory object caching system, generic in nature, but intended for use in speeding up dynamic web applications by alleviating database load.
aiCache is a high performance caching software placed in front of traditional web servers. aiCache intercepts and responds to user requests before they hit origin web servers, App servers or Database servers.
aiCache does work miracles with HTTP, what about HTTPS?
Starting with Version Six, aiCache offers full support for HTTPS, including caching of HTTPS responses.
Why aiCache when we have [insert a proxy web server here]?
aiCache gives you complete control over what and how you want to cache - configured via a single source - no changes to code or web servers or your network setup. It uses RAM to cache response - assuring near 0 time-to-first-byte latency for cached responses. It is extremely light weight and efficient, right-threaded application capable of serving tens of thousands of connected users off a single server and up to 45 000 req/sec, per server, in well-configured setup[1] - that is in excess of 160M req/hr! It collects and makes available, in real time, a wealth of performance data - allowing for dramatically lower TTR when you do have a problem with your web servers or app servers or databases. A site fallback mode keeps your site up even in case of catastrophic backend failure. It allows you to configure cache-by-path feature, selectively bust query parameters, supports cache-busting cookies, expiry of content by response header, comprehensive logging control, SNMP integration , a powerful CLI interface etc. Too many features to mention. It's developed and supported by people that were at forefront of Internet's busier sites from early 1990-ies.
What about analytics ?
We most highly recommend use of Google Analytics - through a simple insertion of some javascript into your pages, you automagically gain access to an incredible set of analytics about your site. Use of aiCache doesn't affect GA in any way, as the magic happens on client side.
If, on other hand, you rely on server-side processing (crunching) of log files, there're no changes to the workflow, but you will need to process aiCache log files instead of the ones collected by the origin web servers. And remember, you can use selective log suppression to dramatically reduce the size of the log files by eliminating logging of auxiliary content altogether. Depending on log crunching software you use, you might need to tell aiCache to log requests in "apache" format or aiCache's own "extended" format.
As an independent party, not affiliated in any way, we still recommend you reconsider and use Google Analytics instead of log file processing.
