Designed and built out highly available VMware vSphere environments using VMware ESX. Environments were built to host industrial control systems and platforms responsible for day to day operations in fulfillment centers. Hardware included Dell PowerEdge R710 servers, Dell PowerVault MD1220i and MD3860i SAN’s, and Juniper EX4550 switches.

Built out an internal DNS system for Blue Apron. Prior to the implementation Blue Apron did not have internal DNS and relied on IP addresses. I choose to use CentOS as the operating system, BIND as the DNS server, and Chef coupled with Chef Server for configuration management and deployment. Internal DNS is currently in use at all office locations with plans to setup another implementation to separate corporate servers from plant servers.

With our growing infrastructure and systems we needed a central logging endpoint that facilitated easy searching, alerting, and scalability. Remote collectors were setup for network segments that do not have an internal route to the main indexer This ensures all devices can log to Splunk in a secure manner (SSL). All of our systems use Splunk for logging and with the use of simple keyword searches, regex statements, and metrics we can easily create dashboards and alerts.

With our growing list of domains and systems, SSL certificate management was non existent. This was a threat as we didn’t know what systems were using what certificate nor tracking expiration dates on those certificates which could cause outages. With Lemur I was able to import all certificates (including expired), setup expiration alerts, note the relationships between the system and the certificate. Having all the certificates stored in a central repository we could easily delegate access to the teams that needed access in order to secure their systems.

Some of the systems that we use do not allow for easy API integration due to technical limitations (OAuth, package dependencies, and language limitations). To overcome these limitations, I created a restful API using Python and Flask to act as a single API endpoint for all of the API’s we wish to interact with. Having a single API endpoint allows us to easily tie in new services with minimal workflow complexity. The tool has been adopted and being further developed by our DevOps team.

Being a heavy consumer of SaaS services, we begin to face scaling issues with user management. I am currently in the process of tying our SaaS services into our identity management platforms for improved user exprience, easier user management, and greater secuirty.