hate these ads?, log in or register to hide them
Page 392 of 392 FirstFirst ... 292342382389390391392
Results 7,821 to 7,826 of 7826

Thread: I don't need sanity, work in IT (thread) every day

  1. #7821

    Join Date
    April 13, 2011
    Posts
    4,535
    Hah. What with? Pure curiosity. I'm a hadoop bod by trade so see very little but kafka.

  2. #7822
    Donor erichkknaar's Avatar
    Join Date
    April 9, 2011
    Posts
    6,445
    Quote Originally Posted by elmicker View Post
    Hah. What with? Pure curiosity. I'm a hadoop bod by trade so see very little but kafka.
    A little sumtin' sumtin' I will most likely open source. I mentioned in a day thread I'm founding this year, this will be one of my gifts to the community and a part on the road to a product.

    Node based with pub sub and a layer based on zmq. I've done a fair bit of architecture on top of, and directed care and feeding of the Kafka ecosystem for a while, and we used to heavily leverage Twisted, which is basically what node is. It will have a redundant in memory substrate that lives with each node instance (so docker container, VM, whatevs) and basically act like a local queue/pubsub mesh but the messages persist in windows the same way they do with Kafka. Some nodes have disk responsibilities. Consumers will be easy to write and it will speak protobuf at the lowest level. Its kind of more something that was designed to run on k8r clusters as a first choice. In truth, we will feed big query with it, at least in my larger architecture, but drivers for hdfs, etc, exist in node land so I hope open sourcing it will encourage those kind of drivers.

    I've worked on stuff like this for a long time and even have a few patents in this domain for the old company.

    Node is the first implementation, Go will follow.
    meh

  3. #7823
    Donor Eard's Avatar
    Join Date
    April 9, 2011
    Location
    Portland. The weird one.
    Posts
    1,927
    Going through similar discussions at work now for our next gen stuff in AWS. Currently have some use of sumologic but ofc sticker shock at our customer instance volume so we'll build our own (ES) and then fail to properly admin it like everything else. In 3 years everyone who worked on it will be gone and we'll have another orphaned critical tool floating around. I run the command center so I'm just forever annoyed at not having every bit of data at my team's hands to do initial troubleshooting and/or not even getting the data because the monitoring/tooling broke and no one is watching.

  4. #7824
    Donor erichkknaar's Avatar
    Join Date
    April 9, 2011
    Posts
    6,445
    Quote Originally Posted by Eard View Post
    Going through similar discussions at work now for our next gen stuff in AWS. Currently have some use of sumologic but ofc sticker shock at our customer instance volume so we'll build our own (ES) and then fail to properly admin it like everything else. In 3 years everyone who worked on it will be gone and we'll have another orphaned critical tool floating around. I run the command center so I'm just forever annoyed at not having every bit of data at my team's hands to do initial troubleshooting and/or not even getting the data because the monitoring/tooling broke and no one is watching.
    I really like prometheus.io for monitoring at the moment. It's going to stay open source and not turn into some weird "open core" thing, which is what the last opensource monitoring tool that showed promise, Sensu, did.

    Prometheus has a very clever architecture that's actually dirt simple to run. I have not tested it at scale, but plan to use it for my next project.

    If you are the guy in the NOC or whatever, you can't do your job without that. You really should own that part, meaning the dev team who will do it.
    meh

  5. #7825
    Donor Eard's Avatar
    Join Date
    April 9, 2011
    Location
    Portland. The weird one.
    Posts
    1,927
    We've got New Relic deployed on all external instances, which while not really providing the level of customer experience monitoring/trending I was hoping for, could certainly be worse. And yeah, we own the deployment and admin of that, although I have a team of 1 dedicated to it. The rest work it on the side of primary alert handling.
    Last edited by Eard; March 26 2017 at 03:09:34 AM.

  6. #7826
    Tailn's Avatar
    Join Date
    April 11, 2011
    Location
    To close to London
    Posts
    4,821
    Comments on Splunk made me wonder what you guys think about Netwrix, this is primarily for auditing ('monitoring') user activity / usage rather than change control (this is a bonus for me) and if there are any suggested Better / cheaper alternates.
    Business worries (MD) are staff absconding with lots of data to competitors primarily so file access trends and alerting etc. are top of the pile.
    Last edited by Tailn; March 27 2017 at 12:09:38 AM.

    "Kerning is serious business"
    And having an image that does not cause Autism attacks even more so.

Bookmarks

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •