Hardware Recommendations

The Sun Startup Essentials program is geared to address some of the most common types of work that today's start-ups are running. Depending on the nature and volume of data, and amount of processing, different types of work can be combined to run together on one server.

As a general guideline, Sun recommends looking at the following hardware platforms for use with each type of work.

Delivery / UI
Type of Work
Description
Requirements
Recommendation
Dynamic web processing
Running a portal, web application with real-time data interactivity; such as: blogs, wikis, tagging, mashups, etc.

Example:
  • Apache (PHP)
  • TomCat
  • JBOSS
  • Sun Java Application Server
  • Horizontally scalable
  • High CPU utilization
  • Medium memory utilization (due to horizontal scaling)
  • Low storage requirements
  • High network traffic
  • Moderate thread concurrency
  • Application layer may have more configuration needs and complexity; suggests higher need for redundancy or larger capacity
  • Sun Fire T1000
  • Sun Fire T2000
  • Sun Fire X2200 M2
  • Sun Fire X4200
  • Static web processing
    Responsible for serving out the static or cached information. Not a huge assembly system - takes hits from users and passes them off to other parts of the systems.

    Example:
    • Apache
    • Sun Java Web Server
  • Horizontally scalable
  • Medium CPU utilization
  • Medium memory utilization (for caching)
  • Moderate storage requirements
  • High network traffic
  • High thread concurrency
  • Redundancy not as important as the layer is replicated
  • Lower support costs from fewer boxes
  • Sun Fire T1000
  • Sun Fire X2200 M2
  • Sun Fire X2100 M2
  • Distributed memory caching and proxy server
    Caching mechanism for objects, web pages, etc.

    Example:
    • Memcached
    • Apache Proxy
  • Horizontally scalable
  • Low CPU utilization
  • Moderate memory utilization
  • Cache servers can be replicated
  • Sun Fire X2100 M2
  •  
     
     
    Copyright 1994-2007 Sun Microsystems, Inc. Close