By Jake Mason
Microsoft Fabric, officially launched on 24th May, is a new Software as a Service (SaaS) end-to-end analytics platform, bundling together Power BI, Data Factory, Synapse Real-Time Analytics, Data Engineering, Data Science and Data Warehousing all into one place. Microsoft is positioning Fabric as a single unified solution for all analytics in an organisation, but more prominently for AI workloads.
In this article we explore 6 key takeaways to understand Microsoft Fabric.
Microsoft Fabric is being presented and sold as a single integrated product with separate applications within. This has many benefits from each application being seamlessly connected together, with OneLake storing all data from each application in one place . From an individual user’s perspective, this means that Microsoft Fabric is a single application with one login experience, one user interface, one storage layer, one permission model and one capacity-based bill.
For the wider organisation, Microsoft Fabric should facilitate better collaboration across all data personas such as data engineers, data analysts, data scientists and business users.
Microsoft Fabric is accessible through a browser at https://app.fabric.microsoft.com. Here you can access each application within Fabric with ease and each application feels well integrated and uniform. This also allows single sign-on within your organisation including 2 factor authentication.
Microsoft Fabric is built upon OneLake, referred to as the “OneDrive for your Data”. So just as you’re used to Microsoft Office 365 products such as Word or Excel saving files into OneDrive, OneLake will store all of your data for each application within Fabric in a single place. This means that there is one copy of your data and one format for your data (Delta – Parquet). This is intended to minimise duplication of data and will make it easier to ensure that this same data is being used across the entire data pipeline from developer to consumer.
Overall, this is a big deal!
Various flavours of languages and expressions (Spark/PySpark for Data Engineering, T-SQL for Data Warehousing, KQL for Real-Time Analytics) can then be used to access your data from OneLake in any application.
Since Microsoft Fabric is a single product, the security and governance surrounding it is also centralised within Fabric Admin Portal, so setting up permissions and adjusting settings will be easier than having multiple separate systems. Settings within Fabric Admin Portal can be adjusted at Tenant level (affecting all Capacities) and Capacity Level (affecting this capacity and workspaces within). You will be able to set different personas for roles within your business so, for example, the data engineers see something different to the data analysts. It is very possible that your company has just one capacity within the tenant, which further streamlines these options.
There is also an Admin Monitoring workspace which comes pre-populated with a report designed to monitor and analyse usage of the platform.
Here is a link to a demo by Microsoft from their Launch Event:
https://www.youtube.com/embed/_Y-XyCRE6ec?start=2775&end=3636
Microsoft Fabric has been fully integrated with Copilot. Copilot can suggest and generate code, suggest next steps and be used as a plug-in to allow a Q&A experience for your users to quickly and easily interrogate data. This is very exciting although it is important to remember that this will be dependent on the quality of the data available. Bad data will give you bad results and conversely good data will give you good results – even some results that you may not have realized without the use of AI.
Here is a link to the demo on Copilot by Microsoft from their Launch Event:
https://www.youtube.com/embed/1o_QDFq6gzE?start=5401&end=6559
Microsoft Fabric is going to have two distinct pricing options for buying ‘capacity’ licenses (F SKUs). You can think of capacity as the amount of compute available for your organisation to run workloads in Fabric. Microsoft has implemented a capacity licence approach based on a sliding compute scale (aka F SKUs). Firstly, a Pay as You Go (PAYG) option which should be available soon near the beginning of June 2023 and secondly a reserved instance yearly option available later this year.
We will need to see how what kind of workloads and processes an F2 SKU will be able to support, but clearly this is a low price point to encourage new users.
Storage in OneLake has a separate pricing meter with storage prices being in line with the currently available Azure Data Lake Storage Gen2 (ADLS gen 2). This means you are getting the OneLake functionality for no additional cost!
Power BI developers and consumers will still need a Power BI Pro licence (US$10 per month). However, with an F64 SKU or higher report consumers do not require a Pro licence. This F64 SKU is equivalent to the Power BI Premium Capacity P1 SKU. An F64 SKU is a much more significant cost (US$11.52 per hour or US$8,409.60 per month).
As mentioned above, Microsoft Fabric is in preview and available as a 60 day trial for all organisations. If you already have a PowerBI Premium licence, you can switch to Microsoft Fabric immediately.
Our expert team are available to demonstrate the capabilities of Microsoft Fabric, and provide guidance on how this solution could cater to your specific business needs and strategic goals.
Join us for a 1-hour briefing call to learn more about Microsoft Fabric from Katalyze Data’s experts.