OmniNet Classic or BYOG Throughput - Speedtest results low

Having the following information can help you narrow down where the issue may be

A. What kind of connection is this? Cable/DSL/Fiber?***
B. Who is the carrier? (Helps you understand if there may be any known carrier issues)
C. What speeds were you seeing before?

1. Used wired ports on Cloud-Link or BYOG (bypassing switches if possible) to establish baseline
- Choose speed testing servers regionally near the data center your Cloud-Link or BYOG is connected to
- Check with and
- Review results

2. Test speeds without the Cloud-Link or BYOG tunnel in place
- Choose speed testing servers regionally near the data center your Cloud-Link or BYOG is connected to
- Check with and
- Review results


***If location is on DSL please read the following article:

Finding the Maximum MTU Value

Below are the kind of considerations we review when speed issues are reported.

----Throughput performance Issues and considerations----

- Testing to carrier related speedtest site would likely yield better results for traffic originating from their own network (Less hops etc.)

- Regular business hours traffic - load from day to day traffic will affect the speedtests since bandwidth is being consumed on the line

- Customer not on a business class ISP subscription - Residential subscriptions do not offer the level of "guaranteed" bandwidth expectations a business plan would carry. This is carrier based behavior, and is NOT uncommon in the industry.

- Machine the test is running on - Testing could be influenced by the machine the test is running from including network card reliability and performance itself, patch cable to switch, switch load, resource availability on the machine

- Replication/Backup operations - Scheduled or Ad-hoc large data transfers will hog up bandwidth and can affect the speedtest results

- Carrier throughput issues - at times carriers don't provide the advertised speed consistently 24/7

- Speed testing to sites not regionally located near the datacenter - Traffic beyond the datacenter traversing networks across states may pass many more hops and be at the mercy of "normal" or expected speed reduction across large regions (i.e. Atlanta to Los Anglees etc.)

- Differing speedtest algorithms - used by different websites that offer speed testing. This is why comparisons are beneficial

- Chatty or physical traffic issues on the local network - If a network switch interfaces devices in between OmniNet and the rest of the segment, speeds are influenced by the network switch performance in between OmniNet and local load on the switch itself (See CPU/Memory/other resource limitations).

- Path from ISP to hosting center - If routers that manage traffic between ISP and our datacenter have less than ideal performance or experience performance degradation

- Issues with the Cloud Link, BYOG device itself or patch cables leading to it

For the more paranoid---

- Carrier analysis - Major ISP's have been suspected to regulate or even capture traffic and analyze traffic that they don't "understand" and this scenario would most likely include encrypted tunneled traffic - Carriers would not disclose this information to the public for obvious reasons

To help us reach a valid comparison, when customer or partner provides speedtest results, there are a couple of things that will help such as:

- Location ISP type and what carrier

- Time/Date of testing

- Service they are testing with ( or etc.)

- Location they are testing against (This should be somewhere regionally near the Datacenter the customer is assigned to)

- A comparison on speed tests from multiple speedtesting sites would be far more valuable

Typically providing this information to OmniNet support will help us quickly reference the service, time, and location to help us understand where the issue may be.

Physical device limitations:

- The Lite/Standard and Standard Plus Cloud-Links are equipped with 10Mb/100Mb interfaces. Throughput for protocols when accounting for overhead will not reach 100 Mbps but rather a percentage of that will be used for protocol overhead etc as is typical with TCP based switching.

- Cloud-Links have been tested and WAN performance does not typically exceed 70-75Mbps while testing when there is no regular business traffic load on the line. (Note: We have observed some locations achieve high 80-90Mbps using the standard Cloud-Link - The 4 LAN port model)

- The Pro Cloud-Link model has been tested to reach consistent WAN speeds of 120 Mbps and is equipped with 10Mb/100Mb/1Gbps interfaces. (Some partners have reported higher WAN speeds)

- BYOG performance varies so you must refer to your manufacturer specs sheets for site-to-site IPSec tunnel throughput.  Some BYOG devices include IPSec offload processing, see if your model has this and test that feature to see if performance improves.

OmniNet Datacenters:

- Do not have throttling on the interfaces that provide hosting for the OmniNet Cloud Links

- Can be susceptible to hosting center issues or carrier specific outages that are upstream from OmniNet that can affect performance.  Thats why checking with a sites' ISP is so important.

- Have monitoring in place to verify resources are plentiful for normal operations

Have more questions? Submit a request
Powered by Zendesk