Good day to All!
The key components of our network are based on x86 servers.
Most of them are Celeron 807 and Celeron 847. The main load
- Filtering and marking of packages: about 100 rules in each of the positions.
Queue tree, nat, and hotspot also located on these servers - about 300-350 users per server.
Average CPU usage is between 50-60%. Used RouterOS 5.21, has been operating for 3-4 years.
About 250-300 Mbps traffic on each server.
The basis of the question: it is possible to replace these servers with new RB1100AHx2? Do they pull the load?
Has anyone done a comparison of performance x86 and RB with a load of mangles and filters?
RB1100AHx2 against x86. Is it possible to replace?
Last edited by streeter on Sun May 15, 2016 12:11 am, edited 1 time in total.
Re: RB1100AHx2 against x86. Is it possible to replace?
On what parameters i should pay attention to see whether it is possible to replace the servers?
Re: RB1100AHx2 against x86. Is it possible to replace?
Neither decided to share their opinions. Despite more than a hundred hits - no one has shown his thoughts.
Okay, I'll be a pioneer. Experience will highlight after my practical research. Dualhead ironburg already ordered.
Okay, I'll be a pioneer. Experience will highlight after my practical research. Dualhead ironburg already ordered.
-
-
andressis2k
Member Candidate
- Posts: 104
- Joined:
Re: RB1100AHx2 against x86. Is it possible to replace?
New RB1100AHx2? It's a quite old device. Although It's a powerful device, I'll go to CCR series.
I'll choose CCR1016 or greater, and of course with latest RouterOS version
Be careful with the upgrade: data flow has some differences between 5.x and 6.x versions
Regards
I'll choose CCR1016 or greater, and of course with latest RouterOS version
Be careful with the upgrade: data flow has some differences between 5.x and 6.x versions
Regards
Re: RB1100AHx2 against x86. Is it possible to replace?
Day Two - normal flight.
Preliminary results:
Performance RB1100AHh2 approximately equivalent Celeron 847. A bit not up to Celeron 1007U. Measurements are made exclusively in my task (see post 1) and may be different from others. The percentage CPU utilization was selected as an index.
This post is posted as information and does not claim to be a bone of contention.
Preliminary results:
Performance RB1100AHh2 approximately equivalent Celeron 847. A bit not up to Celeron 1007U. Measurements are made exclusively in my task (see post 1) and may be different from others. The percentage CPU utilization was selected as an index.
This post is posted as information and does not claim to be a bone of contention.
Re: RB1100AHx2 against x86. Is it possible to replace?
Performance on the same piece of iron under the ROS 6 higher than a ROS 5. It is not considered in the experiment. So it is difficult to conduct parallels.
p.s. Kangaroo chase to death is easier than to alter QoS from ROS 5 to 6.
p.s. Kangaroo chase to death is easier than to alter QoS from ROS 5 to 6.
Re: RB1100AHx2 against x86. Is it possible to replace?
I did a system upgrade to 6.31 (not the last - I know) on the removed server. Subjectively, the performance increase of about 15%. So comparing the actual previous inaccurate.
Is there RB and x86 performance comparisons somewhere ? Please, point me to the topic.
Is there RB and x86 performance comparisons somewhere ? Please, point me to the topic.
Re:
maybe on x86 or Tile - its had some issues aswell as on PPC, but on MIPS-BE its among 6.5 one of best version for latest Qualcomm chips/SoC.6.31 was not considered to be good version. Use 6.32.4 rather instead. Read some threads here to get more info.
Who is online
Users browsing this forum: No registered users and 10 guests