Upcoming Event: The AI Agent Implementation Framework
Upcoming Event: The AI Agent Implementation Framework
JMARK
Springfield, MO
Managed Services / Architecture & Engineering
Amazon AppStream 2.0 | stream.g5.xlarge GPU Instances | Amazon WorkSpaces | Amazon FSx for Windows File Server | AWS Directory Service | AWS Security Hub
JMARK is a managed services provider delivering IT infrastructure, cloud services, and technology strategy to businesses across multiple industries, including architecture and engineering firms that depend on GPU-intensive design tools. Their challenge was clear: on-premises workstations were expensive, hardware refresh cycles were constant, and the cost model was limiting client growth. Avahi deployed a purpose-built AWS environment combining Amazon AppStream 2.0 and Amazon WorkSpaces, enabling JMARK to deliver AutoCAD, Rhino 3D, Revit, and Enscape to thin clients at approximately $2,948 per month. The result is a fully operational, repeatable cloud desktop offering that JMARK can now sell across their entire architecture and engineering client portfolio.
JMARK is a managed services provider that delivers IT infrastructure, cloud services, and technology strategy to business clients across multiple industries. As an MSP, JMARK’s revenue growth is directly tied to the quality and scalability of the managed services they offer. Their architecture and engineering clients present a specific challenge: these firms depend on GPU-intensive design tools that have historically required expensive, purpose-built workstations, creating hardware cost pressure and limiting JMARK’s ability to serve these clients efficiently at scale.
Architecture and engineering firms run some of the most demanding software in the enterprise space. Applications like AutoCAD, Rhino 3D, Revit, and Enscape require dedicated GPU resources that, in an on-premises model, mean a high-spec workstation for every design professional. For JMARK’s architecture firm clients, this translated into significant upfront hardware investment, ongoing refresh cycles, and limited flexibility when onboarding new staff.
JMARK recognized an opportunity: if they could deliver the same GPU performance through the cloud, their AEC clients could provision thin clients or standard laptops instead of $2,000 to $4,000 workstations per seat. But the economics only work if the cloud-delivered experience is credible and the cost is competitive. JMARK needed to validate this with a real deployment for an existing client before building it into their service catalog.
Without a viable cloud desktop offering, JMARK would continue losing potential managed services revenue to hardware-resell-only engagements, and their architecture firm clients would remain locked into the workstation model with no path to workforce scalability. The target cost ceiling was $4,000 per month for the initial client deployment, the point at which the model becomes financially compelling compared to the on-premises alternative.
AWS provided the only managed cloud desktop stack capable of handling GPU workloads at the application streaming layer without requiring JMARK to manage underlying compute infrastructure. Amazon AppStream 2.0 supports GPU-accelerated instance types, specifically the stream.g5.xlarge family, which deliver the NVIDIA graphics performance required for 3D rendering applications like Rhino 3D and Enscape. No equivalent managed service exists in the market that combines GPU streaming, persistent user profiles, enterprise identity integration, and shared file system connectivity in a single, MSP-deliverable stack.
Amazon WorkSpaces complemented AppStream for non-GPU workloads, and Amazon FSx for Windows File Server provided the shared file access that architecture workflows depend on. AWS Security Hub gave JMARK the visibility needed to identify and remediate security findings across the client’s account from day one. The breadth of AWS managed services meant Avahi could deliver a production-grade, secure environment without standing up or operating any server infrastructure on behalf of JMARK or their clients.
As a Premier-tier AWS Partner with deep delivery expertise across cloud infrastructure and managed services, Avahi brought both the technical depth and hands-on execution speed that JMARK needed to validate this model quickly. Avahi’s familiarity with AWS virtual desktop services, identity architecture, and enterprise security hardening made them a natural fit for an engagement that required navigating multiple layers of technical complexity, including GPU quota escalation and Windows BYOL licensing, without slowing down the delivery timeline.
Avahi designed and deployed a hybrid cloud virtual desktop environment built on two complementary AWS services: Amazon AppStream 2.0 for GPU-accelerated application streaming and Amazon WorkSpaces for standard cloud desktops. The architecture is purpose-fit for an MSP model, and JMARK can replicate and resell it across multiple architecture and engineering clients from a single, documented deployment pattern.
The most technically demanding component was the AppStream GPU layer. Stream.g5.xlarge instances, which provide NVIDIA GPU acceleration, were not initially available in the target region due to quota constraints. Avahi escalated a support case with AWS, obtained quota approval for 10 instances including 2 image builder slots, and proceeded to build the application image. AutoCAD, Rhino 3D, Revit, and the Enscape rendering plugin were installed, configured, and verified on the AppStream image. AppStream persistent user profiles were also configured so that user settings and preferences carry across sessions, eliminating the cold-start friction that undermines cloud desktop adoption.
Windows licensing was the second major blocker. Amazon WorkSpaces with Bring Your Own License (BYOL) eliminates the per-user Microsoft Windows license fee that AWS otherwise charges, but BYOL requires importing a qualifying Windows Enterprise image. This process failed twice with evaluation and consumer-edition ISOs before succeeding with an official VLSC-sourced image. Avahi worked through each iteration, sourced the correct media, and completed the BYOL import, enabling WorkSpaces without the recurring per-user licensing cost.
Amazon FSx for Windows File Server was integrated for shared file access, with a login script that automatically maps the FSx share as a network drive at session start. Printers were deployed automatically via Group Policy, and SyncThing was documented as the bidirectional file sync mechanism between the client’s on-premises file server and the cloud environment. Identity architecture decisions were made deliberately: after evaluating a new VPC and directory configuration, the team determined that keeping users on the existing Active Directory and VPC infrastructure was the right call, avoiding the double-setup burden that would have forced users to reconfigure profiles and preferences mid-project.
Throughout the engagement, Avahi also ran a full remediation pass on AWS Security Hub findings in the client’s account. All critical findings were identified and resolved, with confirmation captured in the Security Hub console. High-priority findings were substantially addressed, and a remediation plan was handed off for the remaining items that required ongoing client coordination.
Technical Environment
JMARK Capability Deliverables
Avahi delivered a fully operational GPU cloud desktop environment that JMARK can now actively sell to their architecture and engineering clients. The solution hit the target economics: AppStream-only delivery for 17 concurrent sessions at 8 hours per day comes in at approximately $2,948 per month, comfortably within the $4,000 ceiling that makes the offering financially competitive with on-premises workstations. BYOL on WorkSpaces eliminates the per-user Windows licensing fee that would otherwise apply under a standard AWS-managed model.
Beyond the technical deployment, the engagement produced the full operational foundation JMARK needed to run and grow this service independently. JMARK now has a documented, repeatable migration methodology and support model they can apply across their entire client base. The RBAC configuration ensures JMARK’s own team can manage and support client environments without relying on Avahi access, which is a prerequisite for scaling an MSP offering. A particularly significant output was the quoting capability: Avahi delivered a comprehensive Customer Migration Quote template and AWS Pricing Calculator Checklist covering every billable resource category across approximately 84 resource types. JMARK can now produce accurate, defensible cost estimates for prospective migration clients without relying on Avahi for pricing support, directly enabling their sales motion for new AWS engagements.
Let’s explore your high-impact AI opportunities together in a complimentary session
