Enhancing web applications with cloud intelligence

Editor's Notes

• #6: Tailwind Traders is a full fledged application. But we want to add more. Imagine being able to enhance and extend your application, in many cases without having to modify any existing code? Let’s take a look at what we can build with Azure services to enhance a mobile app we created to complement the website.
• #9: Microsoft Tech Summit FY17
• #10: Microsoft Tech Summit FY17
• #11: Microsoft Tech Summit FY17
• #12: Microsoft Tech Summit FY17
• #14: With serverless, the server, including hardware, infrastructure and configuration of the operating system are all abstracted away. You don’t even have to worry about a web service because it’s provided as part of the platform. Serverless compute is a fully managed service. Some refer to it as Functions as a Service OS and Framework patching is performed for you There is zero administrative tasks and no need to manage any infrastructure You just deploy your code (function) and it runs Your code runs within seconds and for typically shorter periods of times (minutes vs. hours or days) Serverless compute scales quickly (almost instantly) and vastly Automatically scales within seconds No scale configuration is required (there is no way to configure scale or limits) Scales to match any given workload. Scales from zero to handle tens of thousands concurrent functions invocations within seconds Pay only for the time your code is running Serverless compute reacts to events React, in near real-time, to events and triggers Triggered by virtually any event from both inside and outside of Azure A key differentiator about serverless is micro-billing. Serverless resources are billed only when used. That means that you aren’t paying to anticipate an increase in workload, but only when the workload hits. Serverless scales with your business. Let’s take a closer look at micro-billing.
• #16: NOTE: Be sure to talk about choice of languages here as well Use the empty mod30-demo variant of the function Create an HTTP Trigger (not the WebHook quick start, use "additional templates") Mention function security types and choose anonymous Open application insights -> Live Metrics Stream (collapse outgoing requests and general health) Hit the endpoint (either via Test or copy/paste URL, this is preferred so you can capture the endpoint) and show the live metric Point out the "servers" running Run artillery quick --count 100 -n 100 {endpoint} Show auto-scale working, response times, etc.
• #21: 20
• #22: The first example you saw was functions. Functions are literally code and events. And now I’ve revealed a carefully guarded secret: you know where the functions logo came from!
• #23: Although functions are great because they run at scale and only bill when they are active, the “secret sauce” is in triggers and bindings. This is what made it so easy to access the data from table storage and respond to a web event. Instead of standing up a web server, we simply wrote code that responded to a web trigger and were passed a class with all of the information we needed. Instead of scanning for a connection string, opening a database connection, and navigating to table, our binding passed us a class that was already primed to interact with table storage. There are dozens of triggers and bindings available and you can write your extensions as well.
• #24: Here are some of the scenarios we see serverless being used for. Have you ever had to find a scheduling service or install a special service just to run a job at a regular interval? You can trigger functions with a timer trigger and do things like purge expired web sessions from a table. Another common scenario is to trigger a function based on a blob storage event. For example, an ETL process may upload a CSV file that is then parsed, transformed into database rows and then analyzed with a Power BI dashboard.
• #25: Another scenario might trigger from a photograph that is uploaded and this triggers code that automatically generates thumbnails. The last scenario is my personal favorite. You visit a web page, the page calls a function with your user profile and this gathers all of the data collected to deliver a personalized ad. This is why you do a web search for smart thermostat and suddenly see IoT ads everywhere you visit.
• #26: Important Note: the demo as designed will only work with files using the .jpg (not .jpeg, .png etc.) extension. This should be handled by the app but in case you are testing it manually, keep this in mind. Take a picture in the app and save it Navigate to the wishlist container in the storage account Show the image and copy the full URL to the clipboard Take a picture in the app and save it Navigate to the wishlist container in the storage account Show the image and save the full URL Navigate to the mod30-app function and expand, drill into MakeThumbnailHttp Open the "test" tab and change the body to: { "blob": "{url}" } Run and show the execution Navigate back to the storage and show the thumbnail
• #27: Azure Functions : Dépannage et traçage (de courte durée) dans le portail Application Insights : toutes les fonctionnalités relatives à Web Apps / Web Jobs Analyse de trafic, Live Stream, … Dépannage à distance possible avec Visual Studio (pour le code compilé)
• #28: Logic Apps Traçage détaillé et dépannage inclus dans le portail Intégration avec Log Analytics pour analyse : Solution « Logic Apps Management » Autres services serverless Azure Intégration des journaux d’activité et métriques avec Log Analytics, selon le type de ressource
• #30: In the past, managing events was hard. Some of you have built systems that involved multiple servers, load balancers, clustered message queues and database backends just to ensure that you can manage messages at scale across the enterprise.
• #31: Dealing with your own events infrastructure can be complex and cumbersome. It requires a lot of investment and a lot of work. It can also involve a lot of waste. I know of projects that would add a dozen servers in anticipation of a new customer just in case because they didn’t have a good way of knowing what workloads to expect.
• #32: What would be great is to have one centralized service that handles all of the infrastructure and scale out, so we can focus on what’s unique to our business which is the messages themselves.
• #33: Event grid provides this service. It’s fully managed event routing. It provides near real-time event delivery at scale, and I’ll tell you what “near real-time” and “scale” mean with precise numbers in a moment. Event grid works seamless inside of Azure but also is designed to work with third-party platforms and tools. It provides a backbone for event-driven computing.
• #34: At Tailwind Traders, we don’t have to waste time managing severs and preparing to scale. Instead, we focus on innovation. We pay by event so the service scales with our business. We don’t have to invest in backups and redundant servers because the reliability and performance is provided by event grid. This let’s us focus on scenarios in our apps that would have been delayed waiting on infrastructure. The best part is with event grid we can manage all of our events in one place.
• #35: To better understand event grid, there are three common scenarios it is used. The first is as the messaging framework for serverless apps. Resources can generate events that trigger functions and logic apps. Another scenario is ops automation. For example, creating a resource group can send an event that triggers setting up a standard security policy or populating the resource group with a set of standard assets. Finally, a common scenario is integrations behind apps, both current and legacy.
• #36: Event Grid sits in the middle. It is meant for business logic. What’s unique is that event grid doesn’t store message to be polled, but operates on a push model. It pushes messages out to handlers. It does guarantee delivery, so if a handler is temporarily down it will pick up missed messages when it comes back up.
• #37: Event Grid sits in the middle. It is meant for business logic. What’s unique is that event grid doesn’t store message to be polled, but operates on a push model. It pushes messages out to handlers. It does guarantee delivery, so if a handler is temporarily down it will pick up missed messages when it comes back up.
• #38: Event Grid sits in the middle. It is meant for business logic. What’s unique is that event grid doesn’t store message to be polled, but operates on a push model. It pushes messages out to handlers. It does guarantee delivery, so if a handler is temporarily down it will pick up missed messages when it comes back up.
• #39: All of our events are managed centrally. This includes Azure specific events like uploads to blob storage, changes to subscriptions, etc. but also third-party apps that generate custom events. On the other end, many Azure resources are designed to consume event grid events directly, but it is also possible to send them to a web hook you can build in any language on any platform that is capable of processing web requests.
• #40: This is an example of a subscription. It can listen to events generated in Azure or through a custom channel, then route them to an endpoint. It is also to filter events so handlers only receive events they were designed to process.
• #41: I mentioned earlier I would explain the metrics in more detail. Event grid guarantees delivery in less than a second at the 99th percentile. That is what near real-time means. If you need a millisecond respond then event grid is probably not the right solution but for near real-time message delivery it is perfect. That is because it also guarantees that it can handle 10 million events per second per region, and will guarantee delivery in a 24 hour window. It also provides a 99.99% uptime guarantee, can handle 100 million subscriptions per region and will failover to other regions transparently in the case of an outage.
• #42: We built Tailwind Traders to take advantage of event grid. First, let’s level set on some terms. (Main thing to call out is that unlike a queue that typically removes the item when it is popped by the first consumer, event grid messages are delivered to ALL subscribers.)
• #43: Navigate to storage and show events Navigate to the mod30-app function and expand, drill into MakeThumbnailEventGrid Click "Add event grid subscription" and add the details of the storage account Give it a name like "WishlistSubscription" Topic Type is Storage Accounts Select the mod30demostorage storage account Filter to just the Blob Created event Wait for subscription to confirm Expand the logs and keep those open Upload a new image and show it processed in the event grid Navigate to storage and show the thumbnail
• #45: Logic apps is an integration and workflow engine. It started conceptually with BizTalk server and is built on nearly two decades of experience. Logic Apps is a re-imagining of BizTalk as 100% cloud native on Azure with the goal to enable integrations in days or even hours that used to take weeks or even months.
• #46: Logic Apps have literally hundreds of integrations and support for custom integrations as well. You are able to integrate with Azure resources, database servers, SharePoint installations, Office 365, Dynamics, or third party resources like gmail, Twilio for text messaging or Slack for real-time communication.
• #47: The integrations and workflows are all done within a visual designer that makes it easy to integrate, but also loop and branch on decisions so that you can manage different scenarios and edge cases.
• #48: Logic apps are great at connecting resources across the cloud, but they are also capable of integrating with on-premises resources with the on-premises data gateway. That means you can, for example, kick off a workflow in the cloud that results in the execution of a store procedure on your SQL database tucked away in your corporate data center.
• #49: TBD: Tweet a hash tag that does sentiment analysis and then push notifies the app
• #51: Machine Learning and Artificial Intelligence are popular terms today, but what do they mean? At the end of the day they represent a combination of math, algorithms, and statistical analysis that can drive incredible experiences. Only, you don’t have to be a data scientist to take advantage of this technology. Azure provides many services out of the box that you tap into as easily as calling an API.
• #52: These are just a few of the services that are available in each “pillar” of the Cognitive suite. We’ll focus on computer vision to show you how easy it is to integrate with your application.
• #53: Navigate to the mod30-caption logic app Walk through the various steps and explain how one step feeds into the next with variables Add a step to connect with the UpdateDescription function Set the blob to the URL of the blob and description to the generated caption Enable the logic app Upload a new image and show the automated caption
• #54: We did quite a bit with Tailwind Traders, so here’s a recap of some of the services we tapped into and used to extend the application without modifying any code in the original application. We did this by taking advantage of built-in messaging and events and the ability to seamlessly integrate these services in the cloud.
• #55: And that’s a statement around how we feel about Azure. We want to invent with purpose in how we think about it what we deliver to you but also enable you to invent with purpose. And we do that in some key ways. TRANSITION: I talk about these ways as our promises to you on what Azure will always deliver and why we’ll help you.