Couchbase - NoSQL for you! (SDP 2014)
- 1. © Copyright SELA software & Education Labs Ltd. | 14-18 Baruch Hirsch St Bnei Brak, 51202 Israel | www.selagroup.com SELA DEVELOPER PRACTICE JUNE 29 – JULY 3, 2014 David Ostrovsky - NoSQL for you!
- 2. @DavidOstrovsky Senior Consultant @ Sela Co-author of: “Pro Couchbase Server” – Apress Media 2014 “Essentials of Developing Windows Store Apps Using C#” “Advanced Windows Store App Development Using C#” CodeHardBlog.azurewebsites.net
- 3. Agenda What is Couchbase Server? Why should I care? What is this “document” stuff? How do I use it? Where are my documents? What else can I do with it?
- 4. A little history… Couchbase Server is the progeny of two products: Membase - a distributed in-memory cache Apache CouchDB – a document database In 2011, the companies behind the two products merged to form Couchbase Inc. and created Couchbase Server.
- 5. So, what does it do? Couchbase Server is a document database with an advanced memory caching layer. Stores “documents” in JSON format Indexes the contents of documents Couchbase automatically distributes (“shards”) data across the cluster. Data is replicated multiple times for resilience.
- 6. And how does it work? A Couchbase cluster is made up from multiple similar nodes. Nodes are simply installations of Couchbase Server No special head or configuration nodes Client applications use an SDK to connect to the cluster and exchange data. Official support for: Java, .NET, C, PHP, Python, Ruby and Node.js Community support for many more
- 7. Client Cluster Map Node 1 Bucket1 vBucket 1 Doc 1 vBucket 2 Doc 7 Node 2 Bucket1 vBucket 2 Doc 7 vBucket 3 Doc 13 Node 3 Bucket1 vBucket 3 Doc 13 vBucket 1 Doc 1 Active Replica vBucket 1 Node 1 vBucket 2 Node 2 vBucket 3 Node 3 get("Doc 13") vBucket 1 Node 1 vBucket 2 Node 2 vBucket 3 Node 2
- 8. Documents – what’s up with that? UserId Name Age CityId 01 John 35 002 02 Dave 18 003 { "name": "John", "age": 35, "address": { "city": "New York", "street": "…" } }
- 9. When should I use it? High performance requirements Millions of operations per second Very large-scale systems Sub-millisecond reads Serving ads, storing user profiles, etc. Developer friendliness Flexible schema Seamless upgrade of application versions Simple CRUD Straightforward administration
- 10. Working with documents Documents are accessed by their key (aka. ID) Key-based operations are very fast. On a dedicated cluster, you can expect sub- millisecond responses. Items are stored in memory, then written to disk asynchronously. Items are retrieved from the memory cache.
- 11. CRUD All operations are exposed by the Couchbase client object instance. Add/Set store a document in the database. Get retrieves a document by key. Replace updates an existing document. Bulk versions of operations – depends on SDK.
- 13. What database only has one user? Couchbase supports concurrency through Check-and-Set operations. Every item’s metadata has a CAS value, which gets updated with every change. Optimistic concurrency works by comparing client and server CAS values. Pessimistic concurrency works by locking the key exclusively for a specific session.
- 14. Dude, where is my document? In addition to keys (primary index), Couchbase supports secondary indexes, called Views. Unlike key-based operations, which are served from memory, vies are served from disk. File system caches disk reads.
- 15. MapReduce in 60 seconds Views are defined as a MapReduce function. Map function runs for each document and extracts the data for the index. Reduce is optional and can be used for aggregations, such as grouping by value.
- 16. Demo Using Views
- 17. N1QL – How do you pronounce that? N1QL (aka “Nickel”) is an SQL-like query language for Couchbase. Expanded SQL syntax to account for nested documents, arrays, and cross-bucket joins. Currently in developer preview, coming in a future version.
- 18. Demo Was that SQL I just saw?
- 19. Great, what else you got? Cross-Datacenter Replication (XDCR) Couchbase Mobile A NoSQL database for mobile apps. Can be stand-alone or synchronized with Couchbase Server. ElasticSearch Integration Uses ES for querying and Couchbase for retrieval. Hadoop Connector Uses sqoop to move data between CB and Hadoop.
- 20. Questions