I could have chosen a lot of different platforms to create a chatbot for: Facebook Messenger, Skype, WhatsApp, and more. But I decided to go with Viber.

In the United States, Viber doesn’t seem to have a huge following, but I’ve been using it a lot. It’s a very handy way to chat with my wife, send pictures, funny GIFs, and so on. I find it to be more reliable and faster than SMS, especially for pictures. I wish everyone in my family was using it! It’s also a nice side effect that Viber is a Couchbase customer. They switched from MongoDb to support their growing data needs.

Also, Viber’s REST API is simple and well documented. Between the use of serverless architecture and Viber’s API, I couldn’t believe how fast I went from 0 to chatbot.

First, You’ll need to start by creating a bot in Viber (you’ll need a Viber account at some point). Viber will give you an API key that looks something like 30a6470a1c67d66f-4207550bd0f024fa-c4cacb89afc04094. You’ll use this in the HTTP headers to authenticate to the Viber API.

Next, create a new Azure Functions solution. I’ve previously blogged about Azure Functions with a followup on Lazy Initialization.

I decided to use C# to write my Azure Functions. Unfortunately, there is no .NET SDK for Viber (as far as I know), so I’ll have to use the REST API directly. Not a big deal, I just used RestSharp. But if you prefer NodeJS or Python, Viber has got you covered with SDKs for those languages.

Before you start coding, you’ll need to setup a Webhook. This is simply a way of telling Viber where to send incoming messages. You’ll only need to do this at the beginning. I did this by first deploying a barebones Azure Function that returns a 200. I used Postman to set the initial webhook.

Finally, I setup a Couchbase cluster on Azure. Getting started with Couchbase and Azure is easy and free. (You can even use the "Test Drive" button to get 3 hours of Couchbase Server without expending any Azure credit). I created a single user called "viberchatbot", a bucket called "ViberChatBot", and I loaded the "travel-sample" bucket.

For this application, I wanted to create a chatbot with a little more substance than "Hello, world" and I also wanted to have a little fun. Here are the commands I want my chatbot to understand:

I decided not to use any natural language processing or parsing libraries. I’m just going to use simple if/else statements and some basic string matching. If you are planning to create a robust chatbot with rich capabilities, I definitely recommend checking out libraries and tools like LUIS, wit.ai, NLTK and others.

I started by creating a few C# classes to represent the structure of the data that Viber will be sending to my serverless endpoint.

That’s all the basics. Check out the source code for the complete set of actions/operations that the chatbot can do. To test out the bot, I used my Viber app for Android on my phone (and my wife’s, to make sure it worked when I went public).

Beware: by the time you read this, the chatbot I created will likely be taken offline. Anyone else who creates a "Couchbase Bot" is not me!

Here’s a recap of the benefits of this approach to creating a chatbot:

This was really fun, and I could definitely get carried away playing with this new chatbot. It could analyze images, tell jokes, look up all kinds of information, sell products and services, or any number of useful operations.

I would love to hear what you’re doing with chatbots! Please leave a comment or contact me on Twitter @mgroves.

With serverless, you simply write code (usually in form of functions/methods). You can do so with many popular languages, including C#, JavaScript (Node), Java, and so on. This code is deployed to a cloud provider like Microsoft Azure, Amazon Web Services (AWS), Google Cloud Platform (GCP), and more.

Your code is triggered by events. Events can be as simple as HTTP requests, or they could be many other types of events, depending on what the cloud platform supports. (Azure, for instance, supports Timer, GitHub, etc).

Within the cloud, the servers that execute that code are automatically provisioned (and decommissioned) by the cloud provider on an as-needed basis.

You might also see the terms BaaS (Backend as a Service) or FaaS (Function as a Service). Recently, the meaning of the "serverless" buzz word has been expanded, but this post mainly focuses on FaaS/BaaS cloud services.

There are some similarities between serverless and microservices, but they are not the same thing. Both are approaches to break an application into smaller, independent pieces. They differ in what is deployed and what you manage.

As an example, if you are using a microservice architecture, you might have a "shopping cart" service (in addition to other services like "user profile", "inventory", etc). Here’s a diagram of a very simple microservice.

Notice some properties of the microservice:

Contrast this with a similar set of features, this time created with a serverless approach.

In the serverless architecture,

Shown in the above diagram are only standard HTTP requests. You could also use a Timer event, for instance, to check every 5 minutes to see if there are any abandoned shopping carts.

There are some benefits (and trade-offs) when using a serverless architecture.

One benefit is that scalability is handled by the cloud provider. If demand or usage increases, the cloud provider can compensate by adding more server(s) when necessary.

Another benefit is that costs are tied to usage. If you have a service that is constantly in use, you might not see any benefits. But if you have a service that is sporadically used, then serverless may provide cost savings.

Finally, it’s possible that a serverless architecture can reduce administration costs. You don’t need to wait for a server to be commissioned. This may improve agile iteration if commissioning VMs or servers are time consuming. It can reduce your need for IT operations, at least initially, because there are no servers to deploy, fewer servers to manage/upgrade/etc. All this may lead to improved developer productivity.

It’s important to note that serverless is not a silver bullet. Your application may not be a good fit for this kind of decomposition. Also, if you end up deploying a large number of serverless functions, you will still need operations people to manage, monitor, and test your functions. Definitely check out the benefits and drawbacks in detail at MartinFowler.com.

There are several popular serverless providers:

Couchbase Server has partnered with each of these major cloud providers and can run on any of those platforms. In addition, you can run Couchbase Server across multiple cloud vendors for improved reach, disaster recovery, and diversification. You can also use Couchbase Server in a hybrid-cloud situation (a mix of cloud and on-premises).

This makes Couchbase a great choice when you need a NoSQL document database, no matter your cloud strategy:

No vendor lock in. With Couchbase, you aren’t even locked into the cloud, much less a single cloud vendor. With XDCR, you can go cloud-first, and have an on-premises cluster for disaster recovery, or deploy Couchbase to multiple clouds.

Cloud marketplace offerings. You can get started in minutes: on Microsoft Azure, AWS, or GCP.

Tools for your programming language. Couchbase offers SDKs for Node.js, .NET, Java, PHP, Python, Go, C/C++, as well as community support for many others. No matter your serverless platform or language preferences, Couchbase has you covered. Check out this blog post on Azure Functions with Couchbase Server for a technical intro using .NET/C#/Azure.

Scaling. With Couchbase Server, scaling is easy and efficient. Couchbase’s architecture is designed for scaling, with built-in replication, autosharding, and data distribution. Every node in a Couchbase cluster can do both reads and writes, providing efficient use of computing resources and high availability.

Flexibility of JSON. Many apps can benefit from a flexible schema, even if you’re using a relational database. Check out this whitepaper on why Couchbase is the engagement database that can work alongside your transactional and analytical database to provide an exceptional customer experience.

Serverless takes decomposition of your application back-end one step further.

Storing JSON data in Couchbase Server gives you flexibility with both schema and scaling.

Is serverless right for you? It’s not a silver bullet, but if you’re interested in benefiting from lower costs and easier deployment, we’d be happy to help you create a careful plan and discuss whether or not it’s a right fit for your application. You can contact me by leaving a comment, or finding me on Twitter @mgroves.

I had a hard time finding documentation on whether I could use a static object for reuse between function calls. I suppose I should have experimented, like fellow Microsoft MVP Mark Heath did. Instead, I posed the question to StackOverflow.

In short: yes. A Cluster, instantiated and saved to a static member, can is reusable between function calls. According to Mark’s post above, there’s no guarantee how long this value will survive. But that’s an expected trade-off that you make when going "serverless".

Simply using a static member would work, but it’s not thread-safe. There are a few ways to tackle that issue, but an easy way that’s built right into the .NET framework is to use Lazy Initialization with Lazy<T>.

First, I removed the GetBucket and GetCluster methods. Next, I created a Lazy<IBucket> property to replace them.

I just made a single property for a bucket, since that’s all I need for this example. But if you need to use the cluster, you can easily make that its own Lazy property. (Once you have a cluster, getting a bucket is a relatively cheap operation).

When you instantiate a Lazy<T> object, you supply it with an initialization lambda. That lambda won’t execute until the Value property is actually called for the first time.

For instance, notice the Value between Bucket and GetAsync in the updated version of my Azure Functions:

If that’s the first time Value is used, the cluster will be initialized. Otherwise, it will use the already initialized cluster (try experimenting with a Guid instead of a Bucket).

State can be saved between Azure Function calls by using a static member. Make sure that it’s thread-safe (by using Lazy<T> or something like it). Don’t make any assumptions about how long that object will be around.

Anything else I missed? Are you using Azure Functions with Couchbase? I would love to hear from you. Please leave a comment below or ping me on Twitter @mgroves.

Cloud computing basically means that instead of running applications in your own data center, you run it in someone else’s data center. Why would I do that?

Running a data center is difficult and expensive. You have to purchase hardware, manage upgrades, security, networking, and even stuff like electricity, ventilation, and cooling. For some enterprises, this is either not a big deal or it’s worth the hassle. But for many enterprises, the value that you’re delivering is not in the hardware or the operating system and so on, but in the domain expertise that goes into the software you’re building. Then, cloud appeals to enterprises who would rather someone else handle all that other stuff.

A metaphor that I really like was written up by Albert Barron in a blog post called Pizza as a Service (I especially like the diagram). It makes sense for a pizza company to control the whole stack because pizza making is their core competency. But if pizza making isn’t your job, it makes sense to take another option, like dining out, so you can instead spend your time focusing on what you do best.

This isn’t to say that cloud is always the best solution, but it explains why many companies are choosing to move at least some of their infrastructure and platform to a cloud provider like Microsoft’s Azure.

If you’re on the fence, I recommend at least giving it a try, so that you’re prepared for the day that your CTO comes to you and asks "so what are we doing about the cloud?"

Signing up for Azure is easy.

When I first started with Azure, I was very worried that I’d run up a big tab if I wasn’t careful. With the $200 trial, you won’t get charged until you explicitly tell Microsoft to do so. But years later, after my initial trial, I’ve still never had a problem of an unexpectedly high bill.

If you’ve never used Azure or any cloud computing, now is your chance to get started. I’d love to hear about your experiences with Azure, with Couchbase, and your overall impressions of cloud computing. Please leave a comment below, or talk to me on Twitter @mgroves.

For this blog post, I decided to try Visual Studio Preview.

I did this because there is a handy tool for creating Azure Function projects in Visual Studio.

But it only works for the preview version at this time. You don’t have to use these tools to develop Azure Functions, but it made the process simpler for me.

Once I had this tooling in place, all I had to was to File→New→Project. Then under "Cloud", select "Azure Functions".

Once you do this, you’ll have an empty looking project with a couple of JSON files. Right click on the project, add item, and select "Azure Function".

Next, you’ll need to select what kind of Azure Function you want to create. I chose "HttpTrigger". I also chose "Anonymous" to keep this post simple, but depending on your use case, you may want to require an authentication token. After you do this, a very simple shell of a function will be generated (as a C# class). You can execute this function locally (indeed, that is what the local.settings.json file is for) so you can test it out without deploying to Azure yet.

First, I decided that I wanted two Azure Functions: one to "get" a piece of data by ID, and one to "set" a new piece of given data. I started by defining the shape of my data with a simple C# POCO:

Here is the Azure function that I wrote to "get" that document from Couchbase Server:

Some things to note:

Next, run this function locally, and you’ll get a console screen that looks similar to this:

At the bottom, you’ll notice that it tells you the Azure Function URL(s). Assuming I had a document in Couchbase (I don’t yet), I could create an HTTP request with a tool like Postman to: http://localhost:7071/api/HttpTriggerCsharpGet?id=123456

Currently, if I do that, I’ll get "null" as a response (since I don’t have any validation or error checking code). So let’s move on and create a "Set" function.

The "Set" function will be slightly different. I want document information POSTed to it, and I want it to return a message like "New document inserted with ID 123456".

This function has a similar shape to the Get, but some important things to note:

To create a request in Postman, I’ll use a URL of http://localhost:7071/api/HttpTriggerCsharpSet. In the headers, I’ll set Content-Type to "application/json". Finally, the body will be JSON:

Now, when I POST that to the endpoint, I’ll get a response message of "New document inserted with ID f05ea97e-7c2f-4f88-b72d-19756f6a6f35".

I have glossed over how these functions connect to Couchbase Server.

Previously, I mentioned two methods, GetCluster and GetBucket that will connect to the cluster and bucket, respectively.

At this point, most of this code should be familiar if you’ve used Couchbase Server and the Couchbase .NET SDK before. I’m connecting to a single node cluster, and then connecting a bucket that has a password set (I’m using Couchbase Server 4.6).

But, the important thing to point out here is the use of Configuration.AppSettings. In the local.settings.json file, I’ve added these Couchbase settings to the Value section:

When running Azure Functions locally, this file is used for configuration. I have Couchbase Server running locally with a bucket called "azurefunctions". Anything in "Values" can be accessed via Configuration.AppSettings.

Before deploying the Azure Functions, I’ll need to create a Couchbase Cluster on Azure. This is very easy to do, thanks to Ben Lackey’s great work on the Azure Marketplace. Once that’s deployed, deploying the Azure Functions are also easy, thanks to Visual Studio.

This is my first time trying out Azure Functions. This blog post shows a simple demo, but there are other factors to consider before you start using this in production:

Anything I missed? Any more tips or suggestions to share to make this process easier or better? Please leave a comment below or ping me on Twitter @mgroves.