Load balancer Early Access


#1

Hello community,

I am Pavel, Network product manager at Scaleway.

Some of you were asking about managed Load Balancers and highly available IP addresses. It’s finally here.

Well, almost. We opened it in Early Access mode which means that it’s available for free with the following limitations:

  • Access is provided on demand.
  • There is no commercial support, but the product development team will do their best to help you directly.
  • It’s almost production quality, but we can’t guarantee it officially.
  • Only API is available. Web UI is coming soon.

To check it out, just drop us a ticket: https://cloud.scaleway.com/#/tickets/new or even a private message here. For the old-school folks I have an e-mail address: plunin <> scaleway.com

A little guide and API documentation are here:
http://lb.preview.scaleway.com/

(Yes, the diagram is ugly, I know, we’ve already punished those who drew it)

Any feedback, questions, complains, curses, and invective are welcome here or on the #load-balancer-ea channel at scaleway-community.slack.com. For personal insults please use private messages, here it will probably be deleted by moderators.

Kisses,
Your Scaleway LB team


#2

Hello,

Thanks for introducing this, it’s an awesome feature that makes horizontal scaling a lot easier.

I do have a few questions though:

  • What is this load balancer based on, how does it function behind the scenes? (Hardware based or something like a HAProxy server with a failover IP)
  • Does it have a specific bandwith limit (Mbit/s), a monthly volume throughput limit, or a connections limit?
  • Are there any guidelines to how this service will be priced?
  • Will there be any autoscale support? Or at least some API call that returns the amount of active connections on the load balancer?

I understand if you cannot answer these questions, seeing as it is still in preview but I would appreciate any feedback.


Keep up the good work!


#3

Hello @adm,

Thanks for your interest and questions.

What is this load balancer based on, how does it function behind the scenes? (Hardware based or something like a HAProxy server with a failover IP)

The second option. It’s redundant dedicated per-customer instances with some open-source inside and custom magic to provide the high availability and IP failover.

BTW, hardware load-balancer appliances available on the market are not so much hardware-based, in fact. Most of them are x86 or MIPS CPU-based PC-like machines with custom software. While this can be a good choice for an enterprise/private cloud, most of them are limited in terms of scale and multi-tenancy, so we’ve chosen the other way to go for our public cloud platform.

Does it have a specific bandwith limit (Mbit/s), a monthly volume throughput limit, or a connections limit?

No :wink:

Are there any guidelines to how this service will be priced?

This is a sort of question I can’t currently answer publicly but let me say that it will be competitive in comparison to what is available on the market.

Will there be any autoscale support? Or at least some API call that returns the amount of active connections on the load balancer?

I am not sure for the first commercially available release but this is something we are working on.

Thank you once again for the questions.

Wishing you a merry Christmas and a happy New Year!


#4

Hello,

Is there some news about an official launching of the IP load balancer ?

Thanks,
Emmanuel