Upcoming Webinar! Data Center roadmap and strategy

[#1]

Hey, Data Center community! 

We’re excited to bring you a new Data Center webinar with an overview of what’s new and what’s coming in your favorite Data Center products.

For organizations eagerly awaiting all the details on an upcoming feature, planning their next Data Center upgrade, or simply curious about what lies ahead on the Data Center roadmap, this is a webinar you won't want to miss.

Join us on:

US: April 3rd, 2024 at 11:00 AM PST (Los Angeles)
AU: April 4th, 2024 at 11:00 AM AEST (Sydney)
EU: April 4th, 2024 at 11:00 AM CET (Amsterdam)

Register here! https://www.atlassian.com/webinars/enterprise-cloud/how-were-preparing-data-center-for-the-future

We hope to see you there!

[Joseph Chung Yin]

@Claire Chisholm -

For some reasons, I am getting "server error" when I tried to sign-up.  Any advise?

Best, Joseph

[Claire Chisholm]

Hi @Joseph Chung Yin

Can you give it another try? I'm not getting any issues on my side. It might also be security settings from your organization

[Joseph Chung Yin]

@Claire Chisholm -

My re-submission worked.. Thanks.

Best, Joseph

[Lior Efraim]

hi, new to the community, was this recorded? would love to see this.

or maybe there are plans for another session?

[Claire Chisholm]

Hi @Lior Efraim this session is now available on-demand! You can find it here: https://www.atlassian.com/webinars/enterprise-cloud/how-were-preparing-data-center-for-the-future

[Luis Plaza]

Hi!

are there any billing risks on AWS when using S3 as storage for attachments associated with S3 data retrieval during reindexing in Confluence? What's the recommended class of storage in S3?

I am reading this today https://www.infoq.com/news/2024/05/aws-empty-s3-bucket-billing/ 

Thanks

[Michael Andreacchio]

Hi @Luis Plaza, thanks for your question, I can understand your concerns around receiving a large surprise in fees from AWS hosting so I’ve looked into this a little for you.

The high level about attachment indexing is that only text based docs get their contents indexed (.pdf, .word, .txt, etc…). We'll write to S3 the first time a doc is indexed, and then read once per node in the cluster beyond there. For example, in a 4 node cluster a reindex will write max 1 time, and read 4 times per text based file attachment. Note this doesn’t happen for non-text based file attachments.

The feature only supports S3 Intelligent-Tiering so we’ll need to base cost calculations off that tier, reference. To get an idea of costs I’d recommend using the S3 pricing calculator. you'd base this off the number of objects/attachments you’ve got.

Rough back of napkin exercise:

• 1TB of S3 (attachment data) (23.55 USD)
• 10,000,000 PUT operations (0.000005 USD per request)
• 10,000,000 GET operations (0.0000004 USD per request)

Total for billing period = 77.55 USD

Of course, you’d need to gather a rough understanding of how busy your instance is for read/writes, how often you’d be reindexing, and how many text based docs you’re hosting.

Lastly, I’m happy to say that if the above is still a concern that our Open Search implementation will reduce the number of reads across all nodes during a reindex, and will bring it down to only one read per cluster. If you’re interested in trying it, Open Search is currently in EAP in Confluence 8.9, and will become GA during the 9.x release cycle, reference.

Hope this helps, Cheers!

Michael Andreacchio
Confluence DC Product Management