• Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 392
  • Last Modified:

Backups Slowing Down Entire Network

Hello Experts,

I'm a developer that occasionally has to stick my head in the server room so forgive me if this is a simple question:

We have 4 servers One of which is a Microsoft Storage Server that  houses our backup drives(Shared ESata Drives)

We due nightly full backups (150gigs per machine) which take about 14 hours to complete (Long story but they have to be full backups)

Servers are all connected via gigabit Ethernet to a managed switch(SRW2008) that is also connected to 4 other managed switches (Linksys SRW2024) that support desktops and IP phones.  All servers are dual NIC'd (Currently using only 1) Our switches do support VLANS.  

When backups are running the network slows to about 50% normal speed. Servers are not under heavy CPU utilization during backup.

I'm looking for the best way to be able to perform the above backup procedure without affecting other traffic and ideally speeding up the process so that it came complete during off hours.



1 Solution
You already answered your question. I would put them on a separate vlan, or even a separate switch that is trunked to the other switches.
As for speeding up the backup process. I don't see that being a network issue more so than the fact that you are performing nightly full backups.
Aaron TomoskySD-WAN SimplifiedCommented:
Using the second nic to do backups is one option. Teaming the nics is another.
If the servers are on the same swith the traffic should never leave that switch, if traffic is slowing across that same switch then it may have a problem.
The IT Degree for Career Advancement

Earn your B.S. in Network Operations and Security and become a network and IT security expert. This WGU degree program curriculum was designed with tech-savvy, self-motivated students in mind – allowing you to use your technical expertise, to address real-world business problems.

Hang on.  Slowing down the network is not possible because it is switched.  The traffic only flows directly between the Source and Destination,...it does not flow "everywhere",...this is not a Hubbed network,..it is a Switched Network.  Switches create virtual circuits between the two endpoints, so only the two endpoints and the direct path between them ever see the traffic.   So it is not possible to slow down the whole network.

What can happen is

1. that the two servers (the two endpoints) are slower to respond on the network due to the load on the NICs (the nics are where the load is, not the CPUs).  This would give the impression that the network is bogged down when it really is not the network.

2. Another thing that can happen is the back-plane of the switches along the way can be overloaded.  This is less likely since the Back-planes are very fast compared to the port's speed.

3. a third possibility is that there is more than one switch between the endpoints thereby causing the Backbone link between the two or more Switches to be over loaded.

The first and most direct solution is to make sure that both endpoints are plugged into the same Switch and that the Switch is a high quality Switch with good Back-plane speed.

Using a second network to do backups over is a possible solution, but if not handled correctly it can also be a disaster.  Also if DCs are involved,..DCs should never be multi-homed,...so don't do it with DCs.  DCs should not be doing anything but being DCs anyway and should therefore be smaller machines in terms of Drive capacity,...hence making backups short and fast.  The best way to backup DCs is with directly attached backup devices (like inboard or outboard SCSI Tape Drives, USB Harddrives, etc.).  You not likely to truly restore a DCs from a backup anyway unless all DCs are lost at the same time so only backing up one of them is sufficient..
Disk to disk backups can eat up a lot of bandwidth.  On a 1 Gbps network, the transfers between several servers can easily bog down the network.  Since the servers are all in the same room, it should be easy to add a switch for a dedicated/isolated server/storage network.  If you use a switch and dual port NIC's that support the same fast-etherchannel protocols, then you can connect everything at 2Gbps.

Also, do you know what transfer speeds you're getting on the SATA drives?  The write speed may be as much of a bottleneck as the network connections.
mcunnAuthor Commented:
I was able to relieve the pressure on the network by using the second nic's on a separate subnet using a dedicated switch.

 I think the major issue for the slowdown was that most of the data for the network was having to go though the same switch (The file server and router were on the same switch) so we may have very well been stretching  the limits of the linksys backplane.

Also noticed that Jumbo Frames was turned off on one of the endpoints which may have contributed to the long backup times as well.

Backups are now down to a very acceptable 4 hours with minimal load the the primary subnet although and IO of the Esata drive seems to be the new bottleneck.

Thanks for your help.
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

Join & Write a Comment

Featured Post

WEBINAR: 10 Easy Ways to Lose a Password

Join us on June 27th at 8 am PDT to learn about the methods that hackers use to lift real, working credentials from even the most security-savvy employees. We'll cover the importance of multi-factor authentication and how these solutions can better protect your business!

Tackle projects and never again get stuck behind a technical roadblock.
Join Now