Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

Is there a significant increase in latency when accessing an ORDBMS (namely Oracle 11g) residing in the guest VM (from the host), as opposed to natively?

  • The development environment will be used to develop both web applications, and J2EE solutions.

System:

  • Laptop
  • Core i3 CPU
  • 6GB RAM

I intend on:

  • Using Oracle Virtualbox as the type II hypervisor
  • Having all data-tier functionality within the VM (PL/SQL)
  • Having all Oracle middleware residing on the host system (i.e. Glassfish, JRockit)
  • Having control-tier functionality residing on the host system (Java servlets)

Note: This configuration is only intended for the development system, not the production system.


Surely there will be an increase in latency, but what degree of increase is expected? 15-30ms?

  • The development system has 6GB RAM, and a 5400RPM drive

Since I'm working on a variety of projects which require Debian packages, I cannot migrate completely to a RHEL-based VM for development.


To take this to a higher level of abstraction:

  • Is there a traditionally adopted means for piping output from a given VM (within VirtualBox (as opposed to Xen/KVM)) to the host system?

  • Is a socket connection between the host, and guest required when you're accessing a VM locally?


Thank you

share|improve this question
add comment

1 Answer

up vote 1 down vote accepted
+50

I can't answer all of your questions, as I have not done exactly what you are doing. But I have used VirtualBox VMs on a dev machine to separate out different functions. In my case it was having a server service, lets say an SVN server, be hosted in a VM. Then I used the host presenting an iSCSI volume to the VM which had all of the SVN repo information. The VM would mount the iSCSI volume automatically and begin its service.

I achieved this by adding host-only network adapters to the VM and explicitly telling the VMs to look for iSCSI targets at that ipAddress, and then having the SVN service presented out through a public-facing ipAddress.

I accessed and managed the VMs by SSHing into them individually, since they were started automatically from a startup script and ran headless in CLI mode only.

If you have additional questions as to how I did certain things, let me know, I'll see what I can do.

share|improve this answer
    
Had you experienced any significant increase in latency when dealing with the virtual appliance? What my question is asking is: will the increase in latency be prohibitive? –  Delete Jan 10 '13 at 13:37
    
You also give a guess of 15-30 ms, which makes me think that you are looking for exact numbers. If latency is that important, you would most likely need to run benchmarks and measure the differences on specific hardware. A VM will always have slower than native performance, but non time critical situations I was using it for, this delay was negligible. –  CenterOrbit Jan 10 '13 at 16:49
    
I've run NFS servers, SAMBA servers, SVN servers, apache+php servers, 3Ds Max render nodes, and probably more things all from VMs using oracle virtualbox without noticing any sort of noticeable performance degradation. –  CenterOrbit Jan 10 '13 at 16:52
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.