150 likes | 299 Views
Deniss Gaplevsky System engineer at inbox.lv. The portal inbox.lv is a leading national e-service in Latvia More than 80% Latvian inhabitants use inbox.lv services Since 2007 i nbox.lv is in the Latvian Top 20 Media i nbox.lv takes the 10th place among all the Latvian media
E N D
Deniss Gaplevsky System engineer at inbox.lv
The portal inbox.lv is a leading national e-service in Latvia • More than 80% Latvian inhabitants use inbox.lv services • Since 2007 inbox.lv is in the Latvian Top 20 Media • inbox.lv takes the 10th place among all the Latvian media (based on TNS Latvia)
Services Social networking Lifestyle Info Daily Tools Search
Evolutionof Users Platform evolution
A Platform Evolution • 1998-1999 (1,000 users* ) • 1 server:smtp + filters + web + storage + db • 2000-2001 (10,000 users*) • 4servers: smtp + filters, web , storage, db • 2006-2009 (1,000,000 users*) • 5 clusters, 100servers: • web - 14servers • incoming SMTP– 4servers • outgoing SMTP– 4servers • SQL – 4servers • anti SPAM- 5servers • anti Virus - 6servers • storage array - 37 servers *- active users
Free E-mail Service • More than 600’000 logins per day • More than 1’200’000 incoming messages per day • Up to 200 Mbps of traffic • More than 100 servers • More than 250 TB of stored user data • Up to 7’000’000 ads views per day
Scheme of E-mail Service WEB SMTP POP SQL SpamAssasin User Storage Kaspersky Anti-Virus Kaspersky Anti-Spam
Corporate Policy • Open Source software • Quick adaptation • Simple identification and fixing of bugs • Stable time-proven solutions • Simplicity • Unification & Standardization TEAM MONITORING SOFTWARE HARDWARE
Hardware • Reliability • Efficient energy saving • Cost of ownership • Price • AMD based servers – price, quality & energy saving • Intel, broadcom NICs – reliability & performance • 3ware, Adaptec SATA/SCSI – price & quality • HDD Hitachi – reliability
Clustering Basics • One-type tasks requiring CPU+Memorybalancing ability • One-type servers Advantages • Simple maintainability of clustersstep-by-step manual • Simple horizontal scaling of services • High availability of services
Monitoring7x24 • Two parallel systems • Nagios – availability of services • Internal system – availability & thresholds • Historical data and graphs - only for traffic and data storage. • Two types of failures • routine failures failures of hardware or software. Reboot/removal as quick fix (partly delegated to hosting company) • critical failuresfailures ofdatabases and balancers. Must be immediately fixed by inbox.lv team.
Team • Highly qualified staff • Interchangeability • Separated responsibilities • Documentation
Questions? ThankYou!