The Metawerx hosting platform provides a solid, highly available infrastructure to host your Java website, applications and email.
For true bullet-proof hosting and to maintain our promise of extreme uptime, it is important to have systems in place to provide redundancy at multiple levels.
As part of our mission, we continually improve upon this framework wherever possible, taking lessons learned, hypotheticals and customer suggestions to mitigate any areas which could cause your applications to become unavailable.
|Redundancy and Clustering||All Plans|
|High-Availability Java Cluster
- Powerful multi-server Java Cluster with automatic failover and load distribution, built on top of DRBD, our own rewrite of Heartbeat, our internal monitoring, signalling and logging systems, tried and tested and built exclusively since April 2012 for Java Hosting
- Dual RAID SANs are used for failover and mirrored in real-time using DRBD (ie: your live data is on a minimum of 4 hard disks on 2 separate RAID-arrays)
- Any time a JVM is started, restarted, or found to not be running, servers in the cluster contribute to a vote describing their CPU, I/O and RAM load to determine the best server to run the specified JVM
- Your User-Designed Page or our Default Maintenance Page is displayed during the restart/migration of the JVM using Metawerx EpicFailover
- In the case of complete failure of a Java server, all JVMs from that server are immediately distributed to other available servers
- Fully automatic 6-way failover (6 servers in current Java Cluster)
- Read more about our free HA-Java Cluster ...
|High-Availability Database Cluster
- In the case of complete hardware failure of any primary DB server, another server transparently takes control of database requests within 20 seconds
- Automatic 2-way failover
- Real-time block-level replication using DRBD
- Enabled for MySQL 5.x, MariaDB 5.x, PostGreSQL 9.x, MongoDB 2.x, CouchDB 1.x
|High-Performance Dual-SAN (Storage Area Networks)
- RAID-based SAN for redundancy and performance
- RAID: In the case of a single drive failure, no data is lost
- Metawerx: In case of a single-drive, entire RAID-array, RAID controller or complete SAN failure, no data is lost and there is immediate failover to a 2nd array for continued availability/uptime. The 2nd array can also suffer a drive failure during this time, with no data loss.
|Redundant Mail Systems
- 3 SMTP servers in 2 countries (international redundancy)
- Automatic failover POP3/IMAP servers
- Real-time block-level replication between primary and failover
- If our primary SMTP, POP3 or IMAP servers completely fail, you lose no email and connectivity resumes within 30 seconds
- Additional offsite 3rd POP3/IMAP server synchronized every 2 minutes in case of complete failure/destruction of the primary data center
|Redundant DNS Servers
- 3 DNS servers in 2 countries (international redundancy)
- Every server uses dual power supplies, connected to separate power grids
- Every server has at least two network cards
- If primary monitoring systems fail, they are restarted by the secondary monitor
- A minimum of 14 days of rolling backups of your system is kept on-site
- Offsite backup in case of a complete loss of the primary data center
- Secondary offsite backup to prevent overwriting during backup periods, in case of a complete loss of the primary data center or backup-site during a backup
- Yes - in total that's 4 copies of your current data (RAID x 2, DRBD x 2), daily snapshots of your last 14 days and 2 offsite backups, so 20 copies of your files and database stored on at least 5 separate disk arrays in 3 geographically disparate locations!
|Redundant Secure FTP Servers
- Failover FTP server if our primary server goes down
- Seamless access to your data in case of server, network or storage failure
|Maintenance Page during restarts
- Your most common outages are caused by restarting your JVM
- During any restarts, whether manual or from our monitoring systems, our failover device will immediately react and display a Maintenance Page to your visitors, similar to the page you see during maintenance at Google, Twitter or major banks and airlines
- Your maintenance page can be entirely customised to match your brand
- Increase trust in your company during maintenance, improving visitor retention
- Find out more ...
|Failover to HotSpare/LoadBalanced JVM
- Free failover to a 2nd hot-spare or load-balanced JVM when your primary is offline or during maintenance
- Read more about Custom Multi-JVM Failover ...
- Read more about our free HA-Cluster Failover ...
* Requires a minimum of 2 JVMs
* Extensive options available including Round-Robin DNS, Round-Robin Request or High Availability Transparent Load-Balancing using Session-level failover and Distributable WebApps
* Consultancy available