{"id":661,"date":"2017-10-26T09:23:19","date_gmt":"2017-10-26T17:23:19","guid":{"rendered":"https:\/\/www.kenwalger.com\/blog\/?p=661"},"modified":"2018-01-16T11:05:03","modified_gmt":"2018-01-16T19:05:03","slug":"change-streams-coming-soon-mongodb-3-6","status":"publish","type":"post","link":"https:\/\/www.kenwalger.com\/blog\/nosql\/mongodb\/change-streams-coming-soon-mongodb-3-6\/","title":{"rendered":"Change Streams in MongoDB 3.6"},"content":{"rendered":"

Gone are the days when expectations of information were not instantaneous access to the data. In today’s world, people expect real-time results and information. What’s the moisture content of the soil around grapevine X right\u00a0now<\/strong>? Not yesterday, not five minutes ago, but now. Being notified about constantly changing data points is critical in many of today’s applications. In the latest 3.6 release of MongoDB<\/a> a new feature is being introduced to assist with these real-time expectations: Change Streams.<\/p>\n

Change Streams Defined<\/h3>\n

Change streams are being implemented in the driver with a new aggregation<\/a> operator, $changeStream<\/code> and watch<\/code> method in the API. We can now specify a\u00a0$changeStream<\/code> stage at the beginning of our pipeline and request that notifications are sent for specific changes to a particular collection.<\/p>\n

Example Use Case<\/h3>\n

Let’s take the soil moisture example I stated earlier. Imagine that you have a farm with multiple\u00a0moisture sensors<\/a>\"\"\u00a0or a\u00a0TMP36 Analog temperature<\/a>\u00a0sensor\"\" connected to something like a NodeMCU ESP8266<\/a>\"\"\u00a0or LoPy<\/a>\"\" device. Without going into detail about how to set the IoT network up, for the sake of our use case we can assume that they are sending readings back to a MongoDB server. Perhaps the data is being stored on MongoDB Atlas<\/a> even. I have, however, written about the LoPy microcontroller<\/a> and ESP8266 with temperature<\/a> sensors in the past if you are interested in learning more.<\/p>\n

Anyway, for the sake of argument, let’s say that we need to know when the moisture content of the soil drops below a certain level. Moisture values on similar devices are measured as a raw number between 0-1024. The higher the number, the more moisture in the soil. For our use case, we need to know when the moisture level drops below 350 (too dry).<\/p>\n

Change Stream in action<\/h6>\n

Since these recordings are frequently changing, we can set up a change stream to get notified. It’s fairly straightforward to establish a notification with the watch<\/code> method. We could set something like this up:<\/p>\n

change_stream = client.winery.collection.watch([\n  {'$match': {\n       'operationType': {'$in': ['insert']}\n    }},\n    {'$match': {\n        'moisture_level': {'$lt': 350}\n    }}\n]);\nfor change in change_stream:\n    turn_on_water();\n    # Something else magical<\/pre>\n
Now when we have an insert into the database where the moisture_level<\/code> is below 350, we’ll turn our watering system on along with whatever else needs to be done. We could set up a similar change stream for when the moisture level gets high enough to warrant\u00a0shutting the system off at 450.<\/p>\n
This is but one example of where change streams would be a useful tool. Updating reports dynamically on data changes, publish\/subscribe types of functionality, brokers, and many other use cases are now open to native MongoDB actions.<\/p>\n
Change Steam Details<\/h3>\n
You’ll notice that there is an operationType<\/code> being looked at there. Change streams look at five different events, or Operation Types<\/p>\n