Azure Data Platform Tables - Queues - Blobs Azure SQL Azure Search

1. Accesso ai dati con
la Azure Data
Platform
Luca Di Fino
Windows & Windows Phone developer
dev.luke2375@live.it
@luke2375
#azureit15

2. Microsoft Azure Data Platform
+
Database
aggiuntivi

3. Types of Data You Can Store in Azure
• Relational data (Azure SQL)
• Flat semi-structured data (Azure Storage Tables)
• Files (Azure Storage Blobs)
• Serialized Object Graphs in Binary / Base64 (Azure Storage Blobs)
• Messages between systems or processes (Azure Storage Queues)
• Serialized Object Graphs in JSON (DocumentDb)
• Cache (Redis)
• Application data for PHP apps (MySQL managed by ClearDb)
• NoSQL data (MongoDb managed by MongoLabs)
• Big Data (Microsoft HDInsight

4. Azure Data Platform
Demo

5. What is “NoSQL”?
• Modeled using other mechanisms than tables and
relations
• Key-value stores (Azure Storage Tables)
• Document databases (DocumentDb)
• Graph databases for storing complex relationships
(See: http://bit.do/graph-db Ex. Neo4j)
• “Not Only SQL”
• Vertical vs. Horizontal Scaling

6. Azure Table Storage

7. How Azure Storage Tables are structured

8. Anatomy of an Entity
Required Properties
•PartitionKey
•RowKey
•Timestamp
•Etag (think: optimistic concurrency)

9. Even though Azure Storage Tables
are represented in a tabular format,
they are not to be confused with
RDBMS tables

10. Comparing Azure Storage Tables with Azure SQL
• You cannot design Azure Storage Tables using
relational database theory (Codd, normal forms,
etc.) These serve a different purpose
http://bit.do/azure-tables-vs-sql
Maximum row size
1 MB
Max 255 properties
2 GB
Max 1024 columns
Maximum data size 200 TB per table 150 GB per database
Maximum number of rows
retrieved per query
1,000 Unlimited

11. Utilizing Tables in your application
Azure Storage Client library (wrapper around RESTful
HTTP Azure Management API)
Basic course of events:
• Create a connection to the storage account
• Create a connection to a table (or create it if it doesn’t exist)
• Insert entities (Batch insertion)
• Entity classes (POCOs) must inherit from TableEntity class to be serialized for persistence in
Azure Storage Tables
• Perform some command (update, delete, retrieve / filter)
• Retrieve / filter using TableQuery<T>

12. Storage Tables
Demo

13. Azure Storage Queues

14. Azure Storage Queues
• Messages are inserted into Queues and kept in
order until they are “delivered” (de-queued)
• Simple First-In-First-Out queue in the cloud
• Each queue message can only be 64k (plenty!) but
a queue can hold 200 TB of messages!
• Enables highly scalable applications allowing
multiple queue writers to work with multiple
queue readers.
• Handles unexpected spikes in traffic

15. Anatomy of Messages
• Messages can be strings (UTF-8) or byte arrays
• Typical messages contain:
• some proprietary message format (an XML document,
comma-delimited file, etc.)
• A serialized object or object graph (Ideally both writer and
reader would depend on the same interfaces)
• Messages have an expiration date which is, by
default, 1 week

16. Processing Queues
• To process (read) a queue, a reader will grab a bunch of
messages (max: 32) off a queue in a single request
• These are hidden – not removed – until the reader
deletes (de-queues) them OR the reader times out, at
which point they are un-hidden allowing another reader
to attempt to process them.
• Message contents and timeouts can be modified
• Messages can be “peeked” at, which does not hide them
on the queue (you can also peek at the number of
messages on the queue)

17. Why queues are awesome
• Reduces the possibility that data is lost due to timeouts
to the data store or long running processes
• Allows applications to accept data from a user then
throw it over the wall to the reader (i.e., a web job,
worker role, backend process, etc.)
• The reader will get to it eventually – or to handle more
load, just add more instances of the reader temporarily
• Send messages between disparate systems

18. Storage Queues
Demo

19. Azure Storage Blobs

20. Azure Storage Blobs
• Blob = Binary Large Object
• Many practical uses inside and outside of Azure
(i.e., Virtual Machines, logs, CDN origin, backups,
etc.)

21. Anatomy of a Blob
• Any file type and size can be stored in a blob
• Most of the time, you’ll use “block blobs” – 200 GB
• Need to modify ranges of bytes inside of a blob
frequently? Use “page blobs” – up to 1 TB
• Blobs are accessible via URL format:
http://<storage-account>.blob.core.windows.net/<container>/<blob>
• Blobs can live inside of folders (and sub-folders) of a container

22. Securing Blobs - Management Keys
• Container Access:
• Private => Default, only visible by owner
• Public Blob => Read access for blobs only (not container
stuff)
• Public Container => Full read access for blob and
container
• Shared Access Signature – Restricted rights for a
specific time interval (Actually works on tables and
queues, too)

23. Storage Blobs
Demo

24. Azure SQL Server

25. What is Azure SQL?
Core RDBMS SQL Server functionality in the
cloud. High availability, low learning curve,
inexpensive starting point, elastic capacity,
low management overhead.

26. What you get:
• Tables, Views, Stored Procedures,
Security Roles and Users
• Limited T-SQL functionality
• Automatic backups
• Auditing
• Monitoring
• Geo-replication
• Client tools you already know + online administration

27. Service
Tier
Performance
Level
Common App
Pattern
Performance Business Continuity
Max DB
Size
Transaction Perf.
Objective
DTU PITR DR / GEO-Rep
Basic Basic Small DB, SQL opp 2 GB Reliability / Hr. 5 7 Days DB Copy +
Manual Export
Standard S0
S1
S2
Wrkgp/cloud app,
multiple concurrent
operations
250 GB Reliability / Min. 10
20
50
14 Days DB Copy +
Manual Export
Premium P1
P2
P3
Mission Critical, High
volume, Many
concurrent Users
500 GB Reliability / sec. 100
200
800
35 Days Active Geo-
replication
Scegliere l’edizione corretta di SQL Database
Queste informazioni potrebbero cambiare nel tempo.
http://bit.do/azure-sql-pricing

28. RavenHQ è un database documentale (RavenDB SaaS).
MongoDB è un database documentale (JSON).
Cassandra è un database per grandi quantità di dati con una
eccellente replica.
HBase è un database Big Data (Hadoop) disponibile in HDInsight.
Macchine Virtuali con Oracle sono supportate in Azure.
MySQL è offerto com SaaS dal partner ClearDB.
Ulteriori Database in Azure

29. Azure Search

30. Azure Search
Incorpora una sofisticata esperienza di
ricerca in applicazioni web e mobile,
senza doversi preoccupare delle
complessità di ricerca full-text e senza
dover installare, mantenere o gestire
qualsiasi infrastruttura.

31. Azure Search
Ideale per gli sviluppatori cloud
enterprise, produttori di software sul
cloud, architetti che necessitano di una
soluzione di ricerca completamente
gestita.

32. API HTTP/JSON per la creazione di indici, aggounta di
documenti e ricerca
Parole chiave di ricerca con operatori user-friendly (+, -
, *, “”, etc.)
Evidenza dei risultati
Faceting (istogrammi su range di dati, tipicamente
utilizzati nella navigazione di cataloghi)
Funzionalità di ricerca

33. Suggerimenti (auto-completamento)
Query strutturate e complesse (filter, select, sort)
combinate con la ricerca
Profili con punteggio per modellare la rilevanza dei
risultati di ricerca
Supporto geo-spaziale integrato nei filtri, nel sort e nel
ranking
Funzionalità di Ricerca

34. Documentazione
• Azure Documentation - All languages
• Blob Storage feature guide
• Table Storage feature guide
• Queue Storage feature guide
• SQL Database feature guide

35. SQL Database SQL on IaaS DocumentDB
Search HDInsight
MongoDB, MySQL,
Oracle, Cassandra,
Neo4j e altri
Microsoft Azure Data Platform

36. Highlights
Pick the right storage tool for the job
Azure can do most of it
(or you can host in a VM)
RDBMS vs. NoSQL

37. Feedback & Q&A
http://aka.ms/evalazurebcamp