We can handle the most complex AWS deployments.
We are experts in cloud architecture, with particular depth in the Amazon Web Services (AWS) cloud offerings. We have deployed thousands of cloud servers, services, and serverless solutions for consulting clients for everything from PoC and Dev/QA workloads to mission critical, high availability production environments.
The Kickdrum team includes 17 accredited business and 12 accredited technical professional members along with 2 on-staff AWS Certified Solution Architects.
We helped Mapp transform their infrastructure with AWS, increasing performance, scalability, and availability and reducing total cost of ownership.
Kickdrum helped Mapp migrate their leading email marketing system from 3 different on-premise data centers into AWS to take advantage of cloud economics, increased availability, and horizontally and vertically scalable, elastic infrastructure.
Advanced networking and security
To reduce security risks and threats to the production system, we helped Mapp implement secure access and separation of concerns with carefully architected IAM roles, segregated VPCs, and isolated AWS accounts for production and lower environments.
We utilized AWS Snowball, a site-to-site VPN, AWS Direct Connect and AWS Database Migration Service for the 100+TB data migration and Route53 to reliably serve the 30+ custom DNS zones, all while queuing and connecting up with new mail transfer agents (MTAs) hosted in Rackspace.
Optimized for scalability
We helped Mapp achieve better scalability with a custom-architected mix of AWS services and on-demand infrastructure. We created specially tagged CloudFormation templates to easily stamp out new environments from proof-of-concept through production. The new design leverages EBS-optimized EC2 instances that automatically mount a shared Elastic File System and auto-scale using multiple ELB endpoints. We also migrated from Akamai to CloudFront with multiple distributions to reduce cost.
We decoupled session management from app servers using ElastiCache and added failover to a Job Queue engine using multi-AZ MySQL RDS instances. High availability through redundancy with failover support was achieved by deploying app, script, EFS and database servers into multiple AWS availability zones.
A challenging database migration
It took multiple iterations to find the right approach to migrating the 50TB Oracle production database from its on-premise 160-core DL980 with multiple 10gig network cards backed by an EMC VMax. A simple lift-and-shift would not work since AWS did not yet have instances large enough to accommodate.
The initial data migration was accelerated using 2 AWS Snowballs to physically move a compressed backup from the datacenter to AWS. We added a parallelized incremental sync process with periodic recovery to gather post-backup changes and began architecting a migration that included significant re-partitioning, data retention policies, schema remapping, data validation, standby databases and application changes to support sharding. These migrations were all executed as a sequence of proof-of-concepts to vet tools such as AWS Database Migration Service, Snowball, and additional third party schema conversion and data mapping tools. Although each proof-of-concept yielded valuable results and ways forward, a fortuitous release from AWS provided an optimal path forward for the project.
In the meantime, AWS released the new X1.32xlarge instance type which looked like a good candidate for a much simpler migration. To test, we stood up an X1 along with two additional X1 standbys in continuous recovery. After careful benchmarking, analysis, and validation of synced data, we enabled the new setup and it performed very well. We went on to complete the other data architecture improvements as well, leaving Mapp with a more scalable and available infrastructure than their on-premise version could provide.