Hello RodinHooders, I am new to this site and want to introduce myself and my current work. Please help with your feedback, suggestions and comments. And more importantly I am hoping to get noticed by investors as I am seeking funding.
I am a passionate software developer with several years of experience developing large scale software solutions. Gave up my job to follow my passion and currently I am developing a software solution that will enable software/applications to auto root cause defects in real time. The working prototype has reached a good shape now and I want to demo it to investors. I am describing the problem and my proposed solution below and hoping to get feedback, suggestions and more importantly hoping to get noticed by investors.
Before I get into the details, a word about existing solutions like AppDynamics, new relic and splunk etc. These tools help engineers troubleshoot performance and availability issues and they still need engineer’s experience, skill and time to analyze the problem to get to the root cause. The solution I am proposing will provide root cause for functional defects. It will provide defect’s root cause automatically in real time, as defect manifests in production, in the form of an email/alert message with source code line number. For example in case of java application, class file name and line where execution started going wrong, as shown in the image below. This is one of it’s kind solution which helps reduce MTTR (Mean Time To Repair) and software maintenance costs.
Problem
Identifying and trouble shooting of defects/bugs and applying code fixes is one of the important tasks in software engineering and essential to maintain good quality software. Companies spend a lot of money to keep their software quality at highest levels. Engineers spend a lot of their quality time pouring through log files and trying to reproduce the defect in development environments as they trouble shoot to find the root cause. Mean Time To Repair (MTTR) indicates average amount of time required to repair/fix a defect and companies strive hard to keep MTTR as low as possible. With currently available tools, this troubleshooting is still a manual task, which needs engineer’s experience, skill and time to analyze the problem to get to the root cause. This may take days or weeks causing MTTR to shoot high. Seagence automates the task of identifying and troubleshooting defects in real time reducing MTTR by 90%.
Solution
As depicted in the diagram, Seagence uses an agent called SGAgent that hooks into customer’s production environment (can be DEV or QA also). SGAgent injects byte code into application code which enables it to peek into application’s run time. As application users send requests to server and application processes them, SGAgent logs important events and information about user requests and it’s processing to Log Store.
This whole setup is configured to run in 2 stages. In 1st stage, which is acquire stage, SGAgent is allowed to collect as much data as possible (of requests that are successful) while users are accessing the application. When enough data is collected, we feed this collected data to SG Knowledge Builder. As name suggests this knowledge builder will find patterns in the data and gains insights about each request. These insights are then persisted as knowledge in Knowledge Store.
In 2nd stage, which is monitoring stage, we will cut SG Knowledge Builder from reading Log Store and allow SG Prediction Engine to start predicting. The data that is being written to Log Store by SGAgent (as users accessing the application) is now picked by prediction engine which will use the knowledge gained about the request during 1st stage and delivers a decision. The decision can be that the request got processed successfully needing no further action or it can be that the request got failed and in such case an alert, with information as shown in the gif below, will be raised. The image below shows that execution control is returning from process find form method instead of continuing with next statements in that method. An Engineer looking at the message in the alert will easily figure out the root cause and proceed with code fix, saving his troubleshooting time which is otherwise needed.
Thank you for reading and looking forward to your suggestions and feedback. Also requesting help to connect to investors who would be interested.
Thanks,
Srinivas Bitla
Linkedin: https://www.linkedin.com/in/srinivasbitla/
Email: srinivas.bitla@gmail.com