java - CloudBees Service Level Agreement(s) and Capabilities Service -


i have been comparing java paases , really starting cloudbees. have 1 big concern them, , sla/uptime.

after scouring through of documentation, can find one paper offer on slas states:

if using cloudbees paas without taking advantage of high availability options, cloudbees can offer uptime approaches base uptime sla of infrastructure cloud provider.

as same paper mentions, amazon seems offer 99.95% uptime, , know cloudbees runs - largely - on aws/ec2 instances itself.

so spawns number of closely-related sla questions:

  1. if don't take advantage of "high availability" options, can assume cloudbees doesn't guarantee 99.95%? or there documentation elsewhere state uptime is, , remedies failing meet uptime?
  2. what high availability options talking here? read entire developer docs , never saw ha.
  3. what remedies if partner service (like sendgrid mail, or memcachier caching) goes down? 1 thing gae capabilitiesservice where, before go use email api, or caching api, first check master capabilitiesservice make sure services operating. i'd same cloudbees, seems i'd need build myself. that's fine, not sure if cloudbees offers mechanism (api call, etc.) determine if particular service partner on or offline.

thanks in advance!

  1. cloudbees not offer sla on availability nor remedies in form of credits if particular level of uptime not met in month. afaik common other offerings on aws (e.g., heroku). cloudbees offer standard response-time based slas via support agreement. discussed in white paper reference, employ practices our own usage of aws , external providers has helped isolate our users specific amazon issues.

  2. the availability features can make use of include:

    • using multiple instances (and potentially auto-scale). app instances spread cloudbees across different ec2 instances, can avoid downtime in event of ec2 instance failure.
    • using session store. can share session state in separate tier app instance using our offering or partner offering memcachier.
    • using dedicated servers cloudbees sets in multiple aws availability zones.
    • ensuring database used app set in highly available configuration. example, rds simple use cloudbees , supports standbys , read replicas in multiple azs.
    • using app monitoring solutions partners new relic , appdynamics alert of issues.

    the main point of comment using "high availability options" warn people deploying app on cloudbees not make highly available. if ec2 instance fails underneath single-instance deployment, users experience downtime while our internal machinery redeploys working instance, whereas multi-instance deployment experience slower responses until new instance deployed. single-instance databases without standbys or replicas across azs. while stating blindingly obvious lot of people, might surprised how many people assume magic happening.

  3. good point on capabilitiesservice! have ideas kicking around in area, have on own now.


Comments

Popular posts from this blog

jquery - How can I dynamically add a browser tab? -

node.js - Getting the socket id,user id pair of a logged in user(s) -

keyboard - C++ GetAsyncKeyState alternative -