5 ms·

I have a beefy machine and fast internet. Yet, JIRA is annoyingly slow. Half the time, pages on my bank’s website take forever to load, or don’t load at all. I

by akudha 4y ago

I have a beefy machine and fast internet. Yet, JIRA is annoyingly slow. Half the time, pages on my bank’s website take forever to load, or don’t load at all. I don’t have any numbers, just some anecdotal experience like the above. It is not just speed. The UI of many sites suck too. Amazon, GoDaddy, for example. Then there are ads. We don’t see ads in between content anymore. It is content within ads these days. I am thankful that we can do so much online these days. But the experience could be better

scotty79 4y ago
Isn't Jira server-side rendered?
- p_l 4y ago
  Nope, a lot of JIRA runs in the browser. But JIRA main performance issues in my experience are caused by badly sized servers for the level of use.
SgtBastard 4y ago
Jira is a not very good product, implemented by worse engineers, that happens to be better than its competitors (ironically proving the point I made further up thread).
- thiht 4y ago
  > implemented by worse engineers Jira is a 20 years old piece of software. I'm sure today's engineers do their best with the constraints they have. And I'm sure the 2002 engineers did their best with this period's assumptions and solutions. I hate Jira as a product, but please don't attack people like that, have you never had to work on a product you're not 100% satisfied with? :/
  - SgtBastard 4y ago
    https://www.atlassian.com/engineering/post-incident-review-april-2022-outage https://www.atlassian.com/engineering/post-incident-review-a... I stand by what I said. >I’m sure todays engineers do their best with the constraints they have. There are two real constraints in the software engineering world: time and money (especially when we’re talking about an issue tracker, a wiki and a repo + CI\CD pipeline). That Jira and confluences performance remains … not very good, suggests that the engineers are unwilling to, or unable to advocate for, spending the time and money to address this.
  - nazgul17 4y ago
    Hasn't Jira been rewritten from scratch a few years ago?
krautsourced 4y ago
In the case of Jira, how much of that is related to frontend code, compared to backend code / inefficient queries? I find performance suffers mostly when media assets, advertisments, marketing and tracking scripts etc. come into play. The frontend rendering rarely makes a dent at that point.
- DoesntMatter22 4y ago
  The FE is like 5% of the slowness if that. You have a 2 second query and then 50-100ms to render.
Cthulhu_ 4y ago
JIRA, Amazon, possibly godaddy all predate modern-day tools like React and Angular by at least a decade though; Jira seems to have a custom set, Amazon probably does as well. But in most cases, front-end technology isn't the limiting factor.
agloe_dreams 4y ago
The problem with Jira...is that it isn't for you. No really. It is a B2B company that sells an idea to large orgs. "Know what is happening for everything and where you will be in a month. It is not "Enable your workers to work more efficiently". Jira simply doesn't really care about the average dev or such. In fact, most orgs use such a wildly configured version that it is pretty absurd. Let me give you an example: Most companies have a workflow: To Do > In Progress > Testing > Done. Foundationally, Testing and In Progress are tasks done by different operators. Jira doesn't, out of the box, conceptualize who tested a ticket and who developed it. It is all one field. You don't have seperate pointing for testing and development so you can't track QA capacity (Which is helpful to a team but not to overall planning as it just causes rushed QA). You don't have historical views of prior sprints "What was that Ticket I worked on last sprint?" (because that means nothing for forward velocity other than the final number) I could go on..but it all leads back to an absurd UI focused purely on projects and velocity above helpfulness. (And don't get me started on Jira's absurd take on SCRUM) Honestly, I'm kinda shocked no major software company has not just built their own and knocked Atlassian out of the park. I'm sure tracking velocity of other companies would be highly...helpful for Google/MS.
- KronisLV 4y ago
  > The problem with Jira...is that it isn't for you. It's more or less one of those enterprise software packages that will be depended upon in larger orgs, because "nobody ever got fired for choosing $VENDOR". That sort of software tries to do a bit of everything, provide lots of features and customizability, sometimes at the expense of being more complex, bloated and slow. Most people working on their own projects, when fully in control of the decision of what software they want to pick to best suit their own direct needs, might settle on something like Trello or even Kanboard instead. Personally, I self-host OpenProject because its features are a nice fit for my needs, even if it's still a bit sluggish (Rails app, limited hardware resources). Honestly, the performance of Kanboard was blazing fast when compared to anything out there - which is in no small part thanks to how focused and minimalist it is, which in some cases will be exactly what you'll want. As another good example, I'd also put something like the Oracle database in same category: there will be many projects out there developed around it, as well as lots of knowledge, training, as well as entire professions built around it. But most people won't necessarily reach for something like Oracle XE for their own personal projects, they'll instead pick something like PostgreSQL or MySQL/MariaDB, because both of those are a little bit simpler to get started with and keep running, while still provide sufficient functionality. This is an anecdote, but recently I launched all of the mentioned database options one after another in containers, for some testing. Oracle took around 10-20 minutes to start up an de usable. MySQL took about 5-10 minutes to become available for connections. PostgreSQL needed just 2 minutes. That's even before I get into how much data Oracle created in a container volume meant for data persistence (an order of magnitude more than MySQL or PostgreSQL). I experienced similar behavior when setting up my own trial/testing license of Jira because I needed to do some integration dev work - with the resources given to it, I found it to be pretty sluggish and creating new issues took multiple seconds, which does get annoying when you also want to edit some of the fields, or have to do it often as a part of your work. Though I don't have exact numbers for that, I'm afraid.
DoesntMatter22 4y ago
Most of JIRA being slow is the backend not the FE. I've had the same issues but you can watch the backend take 3 seconds to return. The FE isn't much of a factor there.

by akudha 4y ago

scotty79 4y ago

Isn't Jira server-side rendered?

p_l 4y ago
Nope, a lot of JIRA runs in the browser. But JIRA main performance issues in my experience are caused by badly sized servers for the level of use.

SgtBastard 4y ago

Jira is a not very good product, implemented by worse engineers, that happens to be better than its competitors (ironically proving the point I made further up thread).

thiht 4y ago
> implemented by worse engineers Jira is a 20 years old piece of software. I'm sure today's engineers do their best with the constraints they have. And I'm sure the 2002 engineers did their best with this period's assumptions and solutions. I hate Jira as a product, but please don't attack people like that, have you never had to work on a product you're not 100% satisfied with? :/
- SgtBastard 4y ago
  https://www.atlassian.com/engineering/post-incident-review-april-2022-outage https://www.atlassian.com/engineering/post-incident-review-a... I stand by what I said. >I’m sure todays engineers do their best with the constraints they have. There are two real constraints in the software engineering world: time and money (especially when we’re talking about an issue tracker, a wiki and a repo + CI\CD pipeline). That Jira and confluences performance remains … not very good, suggests that the engineers are unwilling to, or unable to advocate for, spending the time and money to address this.
- nazgul17 4y ago
  Hasn't Jira been rewritten from scratch a few years ago?

krautsourced 4y ago

In the case of Jira, how much of that is related to frontend code, compared to backend code / inefficient queries? I find performance suffers mostly when media assets, advertisments, marketing and tracking scripts etc. come into play. The frontend rendering rarely makes a dent at that point.

DoesntMatter22 4y ago
The FE is like 5% of the slowness if that. You have a 2 second query and then 50-100ms to render.

Cthulhu_ 4y ago

JIRA, Amazon, possibly godaddy all predate modern-day tools like React and Angular by at least a decade though; Jira seems to have a custom set, Amazon probably does as well. But in most cases, front-end technology isn't the limiting factor.

agloe_dreams 4y ago

The problem with Jira...is that it isn't for you. No really. It is a B2B company that sells an idea to large orgs. "Know what is happening for everything and where you will be in a month. It is not "Enable your workers to work more efficiently". Jira simply doesn't really care about the average dev or such. In fact, most orgs use such a wildly configured version that it is pretty absurd. Let me give you an example: Most companies have a workflow: To Do > In Progress > Testing > Done. Foundationally, Testing and In Progress are tasks done by different operators. Jira doesn't, out of the box, conceptualize who tested a ticket and who developed it. It is all one field. You don't have seperate pointing for testing and development so you can't track QA capacity (Which is helpful to a team but not to overall planning as it just causes rushed QA). You don't have historical views of prior sprints "What was that Ticket I worked on last sprint?" (because that means nothing for forward velocity other than the final number) I could go on..but it all leads back to an absurd UI focused purely on projects and velocity above helpfulness. (And don't get me started on Jira's absurd take on SCRUM) Honestly, I'm kinda shocked no major software company has not just built their own and knocked Atlassian out of the park. I'm sure tracking velocity of other companies would be highly...helpful for Google/MS.

KronisLV 4y ago
> The problem with Jira...is that it isn't for you. It's more or less one of those enterprise software packages that will be depended upon in larger orgs, because "nobody ever got fired for choosing $VENDOR". That sort of software tries to do a bit of everything, provide lots of features and customizability, sometimes at the expense of being more complex, bloated and slow. Most people working on their own projects, when fully in control of the decision of what software they want to pick to best suit their own direct needs, might settle on something like Trello or even Kanboard instead. Personally, I self-host OpenProject because its features are a nice fit for my needs, even if it's still a bit sluggish (Rails app, limited hardware resources). Honestly, the performance of Kanboard was blazing fast when compared to anything out there - which is in no small part thanks to how focused and minimalist it is, which in some cases will be exactly what you'll want. As another good example, I'd also put something like the Oracle database in same category: there will be many projects out there developed around it, as well as lots of knowledge, training, as well as entire professions built around it. But most people won't necessarily reach for something like Oracle XE for their own personal projects, they'll instead pick something like PostgreSQL or MySQL/MariaDB, because both of those are a little bit simpler to get started with and keep running, while still provide sufficient functionality. This is an anecdote, but recently I launched all of the mentioned database options one after another in containers, for some testing. Oracle took around 10-20 minutes to start up an de usable. MySQL took about 5-10 minutes to become available for connections. PostgreSQL needed just 2 minutes. That's even before I get into how much data Oracle created in a container volume meant for data persistence (an order of magnitude more than MySQL or PostgreSQL). I experienced similar behavior when setting up my own trial/testing license of Jira because I needed to do some integration dev work - with the resources given to it, I found it to be pretty sluggish and creating new issues took multiple seconds, which does get annoying when you also want to edit some of the fields, or have to do it often as a part of your work. Though I don't have exact numbers for that, I'm afraid.

DoesntMatter22 4y ago

Most of JIRA being slow is the backend not the FE. I've had the same issues but you can watch the backend take 3 seconds to return. The FE isn't much of a factor there.