Webinar Meeting Notes: Head In the Clouds S01E01 - Azure Costs Management & Optimisation

About the Event

Welcome to our first webinar event where we will discuss the activities and processes behind successful cloud cost management.


We work closely with businesses in the public sector to implement and manage cloud environments, there are some key successfully used methods to predicting, measuring and reducing your cloud costs, here we will open that discussion and you'll hopefully leave a wiser person with some new skills and knowledge.

Agenda:

  1. 0800 - 0805: Introductions

  2. 0804 - 0815: Visibility, ensuring you're able to visualise predicted and actual spend

  3. 0815 - 08:30: Accountability, Assigning budget owners, defining reporting and setting up alerts

  4. 0830 - 0840: Optimisation, how to optimise your costs, choose the right plans, changing the size and costs of solutions

  5. 0840 - 0845: Conclusion, conclude the event, saying thank you and starting your day

Meeting Record

Agenda item 1 - Introductions:


The meeting started well with introductions for each attendee explaining their background and experiences, the A4S team introduced the role of the webinar which is to inform and educate attendees and also provide a platform for attendees to share their experiences.

Agenda Item 2 - Visibility, ensuring you're able to visualise predicted and actual spend


The group stepped through the use of Azure Cost Calculator and talked through some key examples where costs can easily change through incorrect choices. The examples used were:

  • An Azure VM and the different levels of performance and availability, we discussed the different disk types. The group discussed the user of cost saving measures such as DevTest subscriptions, Spot Instances and high availability groups.

  • An Azure App Services with App Service Plans were reviewed, the different performance and service levels were compared from free tiers to the most costly isolated tiers and where these might be useful.

  • Azure Site Recovery costs were discussed in the calculator, it was also noted IOPs were not shown under ASR meaning a user could easily miss this part of the solution cost. Also re ASR remember that data ingress does not cost, however replication from a failed over instance back to on-premise is egress hence will cost.

Reserved instances were talked through and we mentioned a recent update from Microsoft that RIs over 1 or 3 year can now be paid for on a monthly basis, the only exposure for cost variation being the exchange rate.


Secondly the group stepped through Azure Cost Analysis tool, the impressively granular reports and visualisations allowed budgets and alerts to be set against any solution or area needed. Actions groups were configured to send SMTP budget alert where needed in a test environment.

Some key gotchas for creating unexpected costs were reviewed and shown below:


Azure VMs:

  • Get these on RIs, after these expected an approximate 50/50 split between the RI cost and running costs including storage, network, backup).

  • Windows SA licensing with Hybrid benefit.

  • Can they be replaced with SaaS solutions?

Backups:

  • Backup frequency and retention, both of which are now highlighted in the impressive Azure Backup Reports blade currently in public preview.

Monitoring & Alerting:

  • Reduce frequency of log collection.

Networking:

  • VNet Gateways not needed

  • Static public IPs that could be replaced with Dynamics located by a dynamic DNS name.


Agenda Item 3 - Accountability, Assigning budget owners, defining reporting and setting up alerts


This section covered both technical and organisational aspects of managing cloud budgets. The points raised are below:

  • Organise your assets into resource groups and assign owners - Use resource groups to create budget boundaries that will provide a logical group of assets generating cost. Resource groups are a useful level at which to group you assets and to assign ownership even though there are other ways of reporting on cost such as tags.

  • Assign a budget to an owner, allow the owner to review current burn rate and predict future burn rate - An owner should be able to review current spend using tools such as cost analysis, and can understand usage by reviewing information held with Log Analytics showing usage statistics such as web server hit count, VM memory and processor usage. By understanding end client demand; solution usage can be predicted from which expected cost can be derived.

  • Assign access controls to an owner using a solution such as PAM so that they can maintain control and governance - If an owner controls budget, then they should be able to control IT admin access and changes, using solution such as PAM this can be achieved without hampering operational activities significantly. Ensure changes are managed through a change board where the owner can sanction any cost impacting configuration changes.

  • Configure monitoring and alerting so the owner can visualise the availability and utilisation of their owned asset - Again; budget and budget alerts are key here.

  • Assist the owner to define budgets in Azure - here!

  • Setup cost alerts - here!

After stepping through technical solutions, we then referred to the human processes and disciplines that include:

  • Taking time in your weekly routine to monitor for recommendations and make those changes where suitable - here!

  • Set up routine processes in calendars for owners to undertake reviews and optimisations.

  • Set up routine processes for an overall owner to monitor and support budget owners.

Agenda Item 4 - Optimisation, how to optimise your costs, choose the right plans, changing the size and costs of solutions


Our lead cloud architect Mike led the optimisation discussion (in between recounting his visits to America and his engagements with Microsoft's and Channel 9's Corey Sanders) which covered the following topics:

  • Identifying wastage.

  • Resources and virtual machine sizes.

  • B-Series virtual machines.

  • Leveraging elastic database.

  • Automation - shutting down VM's.

The core elements of how best to use Azure reserved instances was discussed and how there are better alternatives in some cases such a DevTest, automatic shut down routines, spot VMs and trial licenses to test services.


The need and benefits of continuous monitoring, and how you might do this were covered. Some examples of how DevTest environments with some of our clients was mentioned including the subtle differences between DevTest and normal production subscriptions.


Storage costs were discussed as once RIs are in place; they the other largest generator of cost, there are various ways to optimise the use of storage and ensuring they better align to service need.


An A4S hot topic of Recovery Vault was covered and how storage replication impacts cost, we've also been using the recently public previewed backup reports blade which is particularly impressive and integrates seamlessly with Azure Log Analytics data via diagnostic connections.


Azure Hybrid Benefit licensing was briefly covered as well as license mobility.


We discussed the obvious shift in Microsoft’s approach to how the wanted to push their product to market, rather than pushing software licensing they're now educating partners to deliver ‘best practise solutions’ based on the ever-expanding Azure feature set. Microsoft give you a platform, all the building blocks to deploy an environment to fit any requirement. It is up to you how you use these services, implement security, and essentially design your environment and more importantly unless you do, you will not be eligible for the given SLA’s.


Agenda Item 5 - Conclusion, conclude the event, saying thank you and starting your day


In conclusion this was an exciting and productive event, we thank the attendees for their time and hope they enjoyed the session, we're already looking forward to the next events and hope to learn from and build on our first webinar event.


Thank you

The A4S Team!


Webinar - Azure Cost Management
.pptx
Download PPTX • 873KB