We've designed a Data Architecture for our client on Azure wherein, We ingest the sources into a Raw Layer consisting of a Azure SQL Database. This Azure SQL Database acts as a Source Mirror and Has Near Real time sync enabled.
We also have an ODS Layer which is populated from the Previously mentioned Azure SQL Database (Source Mirror) as per the given Data Model. This Layer should ideally take anywhere between 30mins and 1 Hour to Load.
May I Know How Do I Handle the Concurrent Writes and Reads from the Raw Layer (Azure SQL Database, Source Mirror) ? It Syncs every 5 mins with the Sources but also read every 30mins - 1 Hour for ODS Layer.
I've to Use Azure Data Factory to Implement my Data Loads
Yes, Azure Data Factory platform is best fit for such scenarios. Its a cloud-based ETL and data integration tool that lets you build data-driven processes for managing data transportation and data transformation at scale. You can use Azure Data Factory to design and plan data-driven processes (also known as pipelines) that may consume data from a variety of sources. With data flows or computing services like Azure HDInsight Hadoop, Azure Databricks, and Azure SQL Database, you can design sophisticated ETL processes that convert data graphically.
When using control flow, you can use a GetMetadata activity to get a
list of files in a storage account, then pass that list to a for each
activity with the Sequential flag set to false to process all files
concurrently (in parallel) up to the maximum batch size according to
the activities defined in the for each loop.
Here, is the Microsoft Official Document for the Azure Datafactory Connectors Overview | Docs
I would like some advice/tips about the right technology to select in order to store some forecast data on Azure technologies.
My team and I are scraping some weather forecast data everyday from various sources and store it as is on a Azure File Storage. The files format is "grib2" which is a standard format of weather forecast data.
We are able to extract the data from those "grib2" files using python script running on a Azure VM.
We now have several files that represent hundreds gigabytes of data to store and I'm struggling to find which data store from the Azure technologies suits the best our needs in term of praticity and cost.
We started using "Azure Table Storage" first because it's cheap solution,
but I've read on many posts that it is a bit old and not very adapted to our solution as it for example does not allow more than 1,000 entites per query and no aggregation on data.
I considered using Azure SQL db but it seems that it can become very expensive very fast.
I also considered the Azure Data Lake Storage Gen2 (and HDinsight) technologies but am not very at ease with those blob storages and am not really able to say if it can suit my needs in terms of praticity and if it is "easy to query".
By now we just plan to achieve that :
1) Extract data from grib2 files thanks to a python script running on an Azure VM
2) Insert the transformed data into [Azure storage]
3) Query the [Azure storage] from Azure Machine Learning Service or a local R script (for example)
4) Insert the computed data into [Azure storage]
where [Azure Storage] technology is to determine.
Any help or advice would be much appreciated, thanks.
A couple of things I would see here:
To store the downloaded files in raw format (grib2 in your case), either place them on good ol' Azure Blob Storage. Cheap storage exactly for your needs.
Use Azure Databricks to load the data from the storage account and unpack it into memory. (python or scala)
Load the data in memory - still in Databricks - to run you ML inferencing. You could also use SparkR if you really want to.
Store the computed files in a serving layer. This really depends on what you want to do with it later. Often Azure SQL Database is an obvious choice. There is a native Spark connector which efficiently writes data from Databricks to SQL DB.
In addition to using Databricks as your inferencing environment, it's also a good choice for ML training (e.g. utilizing Azure ML Service).
...Or do i need to add the data into a data lake using data factory first and then use databricks as an ELT?
Depends.
Databricks can connect to datasources and ingest data. However Azure Data Factory(ADF) have more connectors than databricks. So it depends on what you need. If using ADF, you need to land the data somewhere (i.e. Azure storage) so that databricks can pick it up.
Moreover, another main feature of ADF is to orchestrate data movement or activity. Databricks do have Job feature to schedule notebooks or JAR, however it is limited within databricks. If you want to schedule anything outside of databricks (e.g. drop file to SFTP or email on completion or terminate databricks cluster etc...) then ADF is the way to go.
Indeed it depends to the scenario I think. If you have a wide variety of datascources you need to connect to then adf is probably the better option.
If your sources are datafiles (in any format) you could consider using databricks for etl.
I use databricks as a pure etl tool (without adf) by mounting a notebook to a storage container in a blobstorage, take huge xml data from there and write the data to a dataframe in databricks. Then I parse the shape of the dataframe and then writing the data into an azure sql database. Fair to say I’m not really using it for the “e” in etl, as the data has already been extracted from the real source system.
Big advantage is the power you have at your disposal to parse the files.
Best regards.
I was going through the Microsoft documents:
https://learn.microsoft.com/en-us/azure/data-lake-store/data-lake-store-overview
I'm new to Azure Data lake and HDInsight. There is a statement in the URL which tells that
"Azure Data Lake Store can be accessed from Hadoop (available with HDInsight cluster) using the WebHDFS-compatible REST APIs."
As per my initial understanding, Data lake store is a store in which any kind of data can be stored. I think, HDInsight also kind of does the same thing.
My question is what is the difference between Azure Data lake and Azure HDInsight? If HDInsight can be used for file storage or any kind of storage then Why to use Data Lake?It would be great if some one could clarify this in details. Thanks.
The easiest way to think of Data Lake is to think of this large container that has like a real lake with rivers coming into the river you never know where the rivers are coming from (or what "type" of river). Azure Data Lake was introduced to make big data easy for developers, data scientists, and analysts to store data of any size. It removes the complexities of ingesting and storing all your data while making it faster to get up and running with big data. Data Lake is able to stored the mass different types of data (Structured data, unstructured data, log files, real-time, images, etc. ) and to blend that together, to correlate many different data types. The key thing here is as we are moving from traditional way to the modern tools (like Hadoop, Cassandra, NoSQL DB, etc). Azure Data Lake includes three services:
Azure Data Lake Store, a no limits data lake that powers big data
analytics
Azure Data Lake Analytics, a massively parallel on-demand
job service
Azure HDInsight, a full managed Cloud Hadoop and Spark
offering
Azure Data Lake Store is like a cloud-based file service or file system that is pretty much unlimited in size. We can run services on top of the data that's in that store. So you could use Hadoop or Spark in an HDInsight cluster, or you could use the Azure Data Lake analytic service, which is a complement to the Azure Data Lake Store. And what that service will let you do is to run jobs that effectively query the data you have stored in the Azure Data Lake store and generate output results.
In nutshell,
Hdinsight is a managed hadoop service (to provide compute support)
Azure Data lake(ADL) is a managed storage service (to provide large amount of storage support)
(Instead of ADL, you can alternatively choose to use Blobs in HDinsight, but Blobs have some limitations (like file streaming to storage via hdinsight cluster is not supported)
Here is the definition from Azure documentation (below):
Azure uses "decomposed hardware method"
You can relate or assume HDinsight as a Hadoop Cluster, Azure Data lake (ADL) as HDFS. But they are detached.
If you want to relate with AWS, HDInsight is equivalent to EMR and ADL is equivalent to EMRFS or S3
If you terminate the cluster, ADL storage stays with the files stored in it. You can access the storage directly using another service or tool (like Azure Data bricks) or you can create one another hdinsight cluster on top of the data.
Hdinsight access the ADL using adl:// , and hdinsight never
store the file blocks in the nodes (like Hadoop does), rather it has
mappings to storage service.
Azure Data Lake Store, is just that a data store. HDInsight can also do that in the cluster that you spin up. However, when you stop that cluster, the data also goes away.
It is common that customers use either Azure Data Lake Store, or Azure storage to provide permanent storage separate from the cluster (compute) used to process the data.
Guy
HDInsight is the analytics service whereas the Azure Data Lake Storage is the storage service. You most likely need both to have functional analytics cluster.
HDInsight provides the cluster, fully manages the open-source packages for analytics (Hadoop, Spark ...etc), and you set up your cluster to use Azure Data Lake Storage which support HDFS API ( Hadoop FileSystem ) on top of Cloud Storage.
Azure Data Lake Storage Gen2 is what you are supposed to start looking at which merges the benefits of both Azure Storage and ADLS in one service.
ADLS Gen 2 documentation - https://learn.microsoft.com/en-us/azure/storage/data-lake-storage/introduction
Azure Data Lake Analytics provides server less compute while using Azure Data Lake Store for data storage, whereas in HDInsight,we need to specify and design for Compute Virtual Machine nodes as per processing requirements. It may be advantageous for developers to work with server less compute in Azure Data Lake Analytics, as scaling needs of Analytics Job are taken care out of box.
We are planning to do batch processing on a daily basis. We generate 1 GB of CSV files every day and will manually put them into Azure Data Lake Store. I have read the Microsoft Azure documents regarding the batch processing and I have decided to use Spark as to batch processing. My question is that after we transfer the data using RDD/DF what would be the next step? how we can visualize the data? since this process is supposed to be run every day, once the data transformation done using Spark, do we need to push the data to any kind of data store like hive hdfs or cosmos before we could visualize it?
There are several options doing this on Azure. It really depends on your requirements (e.g. number of users, needed visualizations, etc). Examples for doing it:
Running Spark on Azure Databricks, you could use the Notebook capabilities to visualize your data
Use HDInsight with Jupyter or Zeppelin Notebooks
Define Spark tables on Azure Databricks and visualize them with Power BI
Load the data with Azure Data Factory V2 to Azure SQL DB or Azure SQL Data Warehouse and visualize it with Power BI.
For Time-Series-Data you could push the data via Spark to Azure EventHubs (see Example notebook with Eventhubs Sink in the following documentation) and consume it via Azure Time Series Insights. If you have an EventData-Stream this could also replace your batch oriented architecture in the future. Parquet files will be used by Azure Time Series Insights as Long-term Storage (see the following link). For Spark also have a look at Time Series Package which adds some time series capabilities to spark.