Peter Saint-Andre wrote: > Someone pointed me today to http://www.mailradar.com/ -- it seems that > we could build something similar for xmpp deployment, if we had a > critical mass of people who want to help with that. Perhaps we could > re-use xmpp.net for that purpose. Anyone interested?
Those external-probe sites can be misleading. They need to be combined with analytical studies to be effective. Sure, Sendmail is the #1 MTA installed on servers because it's installed on every *nix machine and usually enabled by default. Even if some of these servers are hosting mail accounts, it might only be hosting one or two accounts. Maybe it means that Sendmail is the best MTA for small deployments, but I usually hear of 2 or 3 different MTAs that are considered superior to Sendmail for both small and large deployments. A very large ESP that hosts thousands, or millions, of accounts will only be counted a few times by mailradar because the actual servers are very large, hosting lots of users, and hidden behind load balancers. We host 85,000 active accounts, but mailradar will only count us once, maybe 4 times depending on how they interpret our MX records. That's greatly overshadowed by the hundreds of linux machines running under our faculty/staff desks, which happen to also be running Sendmail. It is essentially impossible to get a server/user/software ratio by doing external probes. An alternative way to get the information is to do surveys. Even surveys are not completely accurate since people can lie up or down depending on their motivations. A telling comparison: Compare the results of this survey-based study with mailradar's results. http://blogs.sun.com/jhawk/entry/ferris_newsletter_sun_email_1 Jesse -- Jesse Thompson Division of Information Technology, University of Wisconsin-Madison Email/IM: [EMAIL PROTECTED]
smime.p7s
Description: S/MIME Cryptographic Signature
