5 ms·

Brightbox announces easy to use cloud load balancers

19 points by thehodge 16y ago

thehodge 16y ago
We've been using them with pizzapowered.com stuff for a while whilst they are tested internally and we've found it really simple to use
wlll 16y ago
I love the speed you can get stuff done compared to EC2.
- ambirex 16y ago
  I'm curious, what stuff are you talking about? Admittedly, ELB's cname restriction is a little of a downer but you could always spin up an instance to act as your own LB. Also, their 3rd point of needing to rebuild an ELB is a little of a drag but it is very trivial to rebuild by hand, even faster if you have scripted it.
  - wlll 16y ago
    Moving EC2 elastic IPs can take a while in my experience, and also time for creating servers can be unpredictable. With the BrightBox stuff "Load Balancers are normally created and handling traffic within 90 seconds.", also the video here http://blog.brightbox.co.uk/posts/announcing-brightbox-cloud-the-uks-first-true-iaas-platform http://blog.brightbox.co.uk/posts/announcing-brightbox-cloud... seems to show server booting and ip assigning as pretty snappy. If they can maintain this responsiveness in the face of lots of production usage then it would be a good selling point.
  - oceanician 16y ago
    "but it is very trivial to rebuild by hand, even faster if you have scripted it" trouble is, anything that is trivial by hand for you, after you've done it 20 times is tricky for Sys Admins to pick up. And if you script it, when an automated solution is already available, you could be using that time to build business logic, which I'm guessing the main point of most people's role here?
tastybites 16y ago
I think I'm detecting a trend. Take an open source project (LVS perhaps, in this case?), scale it onto a bunch of servers running Xen (your own, Amazon's, doesn't really matter), wrap an API around it, and you've got the next cloud service.
- aaronblohowiak 16y ago
  As DevOps becomes more popular, IT becomes more sophisticated or developers care more about the system beyond their application code, I think we will see greater appreciation of the interplay between networking, systems, operations design and application design. Particularly, I hope to see some innovation in layer 6 and 7 load balancing and routing (and growing cultural awareness of the opportunities such things could provide.)
- jeremyjarvis 16y ago
  @tastybites: a /little/ bit more to it than that ;) we do run the loadbalancing layer across multiple locations which is non-trivial. [Disclosure: I'm co-founder at Brightbox]
  - tastybites 16y ago
    As always, the devil's in the details. I wasn't being snarky - it's actually a great way of doing it since you're leveraging so much existing work. I'm also working on project that could be described in the same way.
patrickgzill 16y ago
I wonder about latency issues - how much is added by the typical LB setup in this case?
- jeremyjarvis 16y ago
  Patrick - negligible, it's all on the same network. [Disclosure: I'm co-founder at Brightbox]

19 points by thehodge 16y ago

We've been using them with pizzapowered.com stuff for a while whilst they are tested internally and we've found it really simple to use

wlll 16y ago

I love the speed you can get stuff done compared to EC2.

ambirex 16y ago
I'm curious, what stuff are you talking about? Admittedly, ELB's cname restriction is a little of a downer but you could always spin up an instance to act as your own LB. Also, their 3rd point of needing to rebuild an ELB is a little of a drag but it is very trivial to rebuild by hand, even faster if you have scripted it.
- wlll 16y ago
  Moving EC2 elastic IPs can take a while in my experience, and also time for creating servers can be unpredictable. With the BrightBox stuff "Load Balancers are normally created and handling traffic within 90 seconds.", also the video here http://blog.brightbox.co.uk/posts/announcing-brightbox-cloud-the-uks-first-true-iaas-platform http://blog.brightbox.co.uk/posts/announcing-brightbox-cloud... seems to show server booting and ip assigning as pretty snappy. If they can maintain this responsiveness in the face of lots of production usage then it would be a good selling point.
- oceanician 16y ago
  "but it is very trivial to rebuild by hand, even faster if you have scripted it" trouble is, anything that is trivial by hand for you, after you've done it 20 times is tricky for Sys Admins to pick up. And if you script it, when an automated solution is already available, you could be using that time to build business logic, which I'm guessing the main point of most people's role here?

tastybites 16y ago

I think I'm detecting a trend. Take an open source project (LVS perhaps, in this case?), scale it onto a bunch of servers running Xen (your own, Amazon's, doesn't really matter), wrap an API around it, and you've got the next cloud service.

aaronblohowiak 16y ago
As DevOps becomes more popular, IT becomes more sophisticated or developers care more about the system beyond their application code, I think we will see greater appreciation of the interplay between networking, systems, operations design and application design. Particularly, I hope to see some innovation in layer 6 and 7 load balancing and routing (and growing cultural awareness of the opportunities such things could provide.)
jeremyjarvis 16y ago
@tastybites: a /little/ bit more to it than that ;) we do run the loadbalancing layer across multiple locations which is non-trivial. [Disclosure: I'm co-founder at Brightbox]
- tastybites 16y ago
  As always, the devil's in the details. I wasn't being snarky - it's actually a great way of doing it since you're leveraging so much existing work. I'm also working on project that could be described in the same way.

patrickgzill 16y ago

I wonder about latency issues - how much is added by the typical LB setup in this case?

jeremyjarvis 16y ago
Patrick - negligible, it's all on the same network. [Disclosure: I'm co-founder at Brightbox]