
4
Introduction
This guide provides information on the guidelines and recommendations for networking equipment
and its use with Scale Computing HC3 clusters. This guide covers general concepts rather than specic
conguration options. If you need information on specic conguration information for a switch or other
networking product you can refer to the guides in the Scale Computing Knowledge base or the website
of the manufacturers of the switch or product.
Minimum Switch Attributes
When using switches other than a recommended switch, the following features and attributes should be
strongly considered.
• VLAN support
• Managed switch
• 1 GbE or 10 GbE
• 72/144 mpps for 24/48 port switches respectively
• Allows disabling of spanning tree protocol (STP) on a per port basis
• Supports 802.3x ow control
• (Optional) Stacking or dedicated interconnnect for High Availability (HA)
• (10 GbE Only) SFP+ ports - twinax cables are recommended as they are inexpensive and
compatible
Recommended Switches
These switches are expected to work based on published specications, successful customer
deployments, and/or internal testing. Switches not listed here may work as long as they meet the
stated requirements above in Minimum Switch Attributes. If it is determined that performance
or cluster redundancy issues are due to network latency, dropped packets, or other network issues, a
recommended switch may become the required action from Scale Support.
1 GbE Switches
• HP Procurve 2920 Series
• Datasheet: http://bit.ly/1xcpHkQ
• Dell Networking N2000 Series
• Datasheet: http://bit.ly/1lYBYkn
10 GbE Switch
• Dell Networking N4000 Series
• Datasheet: http://dell.to/Qba0VS
• Mellanox SX1012
• Datasheet: http://bit.ly/12p1Y24
Known Successful Switches (Untested)
These switches have been seen to be successful in various customer deployments. Switches listed here
have not been tested by Scale and Scale does not offer any guidelines for conguration outside what is
listed in this guide in Spanning Tree (STP) and Flow Control. Switches not listed here may work as
long as they meet the stated requirements above in Minimum Switch Attributes. If it is determined
that performance or cluster redundancy issues are due to network latency, dropped packets, or other
network issues, a recommended switch may become the required action from Scale Support.
Komentáře k této Příručce