Dev Dives: Streamline document processing with UiPath Studio Web
Slow is the New Down
1. Slow is the New Down:
Keep Your Business in Business by Ensuring Successful
User Experiences in the Cloud
DevOps Summit: June 2016
2. • About AppNeta and Our Customers
• The Significance of the Cloud
• Moving to the Cloud: Things to Consider
• Impact of Cloud on User Experience and Performance
• In The Cloud: How to Pierce the Cloud and Ensure User
Experience Success
Agenda
4. Significance of the Cloud
Amazon Web Services (AWS) generated $7.88B in revenue with Q4 2015, up 69% over last year
Source: 451 Research Market Monitor
13. Continuous Monitoring is Required
Cloud Deployments are sharing network capacity with thousands of
other users and applications
Ensure that the application delivery path from your users into your
chosen cloud provider has enough reliable capacity
23. Summary: User Experience to the Cloud
Minimize geographic distance between
Users & Cloud Data Centers
Performance into Cloud Providers
varies widely and changes constantly
Slow is the new Down
24. Cloud Network Performance is an Option
Apps are becoming more distributed
in the Cloud and across Clouds & Networks
within Cloud Providers
are extremely complex and congested
Choose a solution with visibility across the
entire App Delivery environment
Just to set some context here for you about AppNeta, we are an application performance monitoring company. We focus on providing a complete APM solution, across the applications that you develop or build internally, the business critical SaaS applications you’re using to run your business and the networks that deliver them.
We’re a 100% SaaS platform with a large, international customer base of amazing companies that use AppNeta to monitor and manage their applications and make sure that the end-user experience is above expectation.
Make sure it meets you Model
IaaS or PaaS?
Public / Hybrid / Private Cloud?
Billing model?
What Features are you looking for:
Autoscaling
CDN
Block Storage
Database as a Service
DNS
Messaging Services
VPN Access
Just to set some context here for you about AppNeta, we are an application performance monitoring company. We focus on providing a complete APM solution, across the applications that you develop or build internally, the business critical SaaS applications you’re using to run your business and the networks that deliver them.
We’re a 100% SaaS platform with a large, international customer base of amazing companies that use AppNeta to monitor and manage their applications and make sure that the end-user experience is above expectation.
Just to set some context here for you about AppNeta, we are an application performance monitoring company. We focus on providing a complete APM solution, across the applications that you develop or build internally, the business critical SaaS applications you’re using to run your business and the networks that deliver them.
We’re a 100% SaaS platform with a large, international customer base of amazing companies that use AppNeta to monitor and manage their applications and make sure that the end-user experience is above expectation.
Just to set some context here for you about AppNeta, we are an application performance monitoring company. We focus on providing a complete APM solution, across the applications that you develop or build internally, the business critical SaaS applications you’re using to run your business and the networks that deliver them.
We’re a 100% SaaS platform with a large, international customer base of amazing companies that use AppNeta to monitor and manage their applications and make sure that the end-user experience is above expectation.
Map
https://www.google.com/maps/d/u/0/edit?mid=zSh1_CbTuJ8Q.kv9cGWf6oKsI
AWS
Current - 13
Coming Soon - 5
Azure
Current - 19
Coming Soon - 4
Google
Current - 4
Coming Soon - Not announced
Azure Locations
https://azure.microsoft.com/en-us/regions/
Google Compute Engine locations:
https://cloud.google.com/about/locations/
As soon as you go to the cloud, every office is a remote office
The application experience of each end user can be impacted by poor performance at any point in the application delivery path, and this performance from your offices, over the internet and into the Cloud changes constantly
This performance should be measured over time before, as the data above for a 1 month period shows that even on the best performing cloud provider, Google Compute Engine, there are huge swings in the achievable performance, a 163% swing from the lowest to the highest network capacity and over 1.2 Gbps in available capacity
https://demo.pathviewcloud.com/pvc/reports/dataPerformanceComparisonReport.html?st=3&reportName=DataPerformanceComparisonReport&pathIds=90781,107109,107057&startDate=1459483200000&endDate=1461331369615
One very surprising observation made during our Cloud service provider investigations is the level of packet loss. Both at the ingress to the target and internal within the Cloud providers we saw meaningful amounts of packet loss, almost 3.5% peak for Microsoft Azure. The overall impact of this packet loss is that during the peak loss periods the effective transaction time of web applications served doubled, taking twice as long for the same pages to be delivered.
When investigated by our Chief Architect
Just to set some context here for you about AppNeta, we are an application performance monitoring company. We focus on providing a complete APM solution, across the applications that you develop or build internally, the business critical SaaS applications you’re using to run your business and the networks that deliver them.
We’re a 100% SaaS platform with a large, international customer base of amazing companies that use AppNeta to monitor and manage their applications and make sure that the end-user experience is above expectation.
99% of the 5 billion API calls Netflix handles per day are internal
Why go to microservices?
Fast deployment: Code changes for each service can be made independently (as long as the API contract with other services isn’t violated) and therefore, build+test+deploy cycles speed up dramatically. Netflix, for example, often deploys code a hundred times in a single day thanks to their early adoption of the microservices architecture!
Efficient scaling: Each microservice can be scaled independently, which is a much more efficient way of scaling an application, because not every part of an application experiences the same amount of load and needs to be scaled equally.
Design autonomy: Developers get the freedom to employ different technologies, frameworks, and design patterns to design and implement each microservice, pursuing a horses for the courses strategy as necessary.
When deploying within a cloud provider you choose the CPU and Memory, even Storage IO of your virtual images but most users just assume that they get a 1Gbps virtual interface on all hosts.
Standard images within AWS vary from 62.5Mbps to 500Mbps
But, all major cloud providers have High Network Performance options, up to 10Gbps
If you have a capacity intensive application, make sure you are choosing the right options.
But Capacity is not the only consideration...
Challenges in Deploying a distributed application in the cloud
This is an AppNeta application hosted in AWS, a production Python application As with a lot of multi-tenant SaaS applications we shard the database for data isolation and also for performance
This Host is currently talking to 144 distinct virtual databases, 11 of which are having packet loss above with the peak at 8.5%
Drilling down into one of these paths it is clear that the network inside of AWS is an extremely complex software defined network, with 19 layer 3 hops from source to destinationAWS is over delivering on the 1Gbps network connection we are paying for, with 2.77Gbps achievable from source to destination, but the issue is that packet loss.TCP based applications will automatically retransmit these lost packets, but that retransmit process has a huge impact on the actual performance of your application to the point that this network performance is automatically used as part of our DevOps “Bad Apple” detection process where hosts are continuously monitored and if any number of specific metrics degrade, load is shifted to other nodes and the offending “Bad Apple” is restarted, resulting in a fresh AMI image.
So the “Have you tried turning it off and on again” approach not only works for your parents wireless access point, it works inside the engine running 35% of the internet.https://appneta.tv.appneta.com/app/tv_prod_web/map#source=tv_prod_web&target=752158396
What are the downsides of this monitoring?
There is a slight amount of overhead, 20 to 50 packets per minute to determine
Once you go to the Cloud, every
Application Delivery Path
Once you go to the Cloud, every
Application Delivery Path