[REQUEST] 2 x Dedicated server + cross connect

MarkTurner

@AnthonySmith - its this server:

http://www.delimitervps.com/landingpage/blades/?p=X515016500&l=Atlanta

So it comes with:

4 Cores - 2 x Xeon X5150
16GB RAM
500GB Disk
5000GB Bandwidth
1 IPv4 IP Address
/64 IPv6 Subnet
Gigabit Port
Hardware RAID controller
Dedicated IP-KVM (HP ILO)
Dedicated Power control
No charge instant OS reloads
Wide choice of OS
Custom ISO install via vMedia

VLAN is for the secondary ethernet to give you a private network on eth1.

AnthonySmith

@MarkTurner I don't want a private VLAN I want a cross connect between servers, is that a problem?

MarkTurner

@AnthonySmith - not possible. We only have 10GE ports on our blade servers, the cost to allocate a dedicated 10GE port would be way more than the cost of the server.

AnthonySmith

@MarkTurner said:
AnthonySmith - not possible. We only have 10GE ports on our blade servers, the cost to allocate a dedicated 10GE port would be way more than the cost of the server.

Shame, I did specify that as a requirement.

MarkTurner

@AnthonySmith - a piece of cable or a dedicated VLAN - there is no difference unless you want to use the link state to signal morse code between the units.

AnthonySmith

@MarkTurner I am looking for as near 0 latency as possible with redundancy from a failed switch or in this case blade chassis, perhaps a blade solution is not the right one for this.

MarkTurner

For almost zero latency - then use Infiniband for the cross-connect

pcan

@AnthonySmith said: perhaps a blade solution is not the right one for this.

A proper blade enclousure has better redundancy than a regular server (usually 6 PSU vs 2; internal ethernet fabric is redundant; the backplane is redundant too). It is more geared towards a high availability solution and (usually) has some fast interconnect such as infiniband.

Multiple blade faults can be triggered by relatively rare issues: PDUs or environement/rack failures; a botched enclousure firmware upgrade (it may force to execute a restart to regain management functions). For a high availability solution, I rather choose a blade than a rack server (at equal conditions).

fileMEDIA

AnthonySmith said: failed switch

You have bigger problems when both servers are using the same switch for the public network. Switch failure, both servers are not available and only internal network works.

You should extend your requirements with a fully redundant network for the public NIC also.