Finch Paper is located at the foothills of the Adirondack Mountains, on the banks of the Hudson River in northern New York. They’ve been making paper here for over 100 years with a keen understanding that the viability of their mill relies on the wise use—and reuse—of resources.
They are an integrated paper mill practicing sustainable forestry, generating their own green power, and making their own pulp to produce high-quality, affordable uncoated printing papers. They personally promise that their paper is made responsibly — with a steadfast commitment to keep forests as forests.
Finch Paper was faced with the tough decision of updating their current server environment while keeping costs down, increasing performance and systems availability for a 24/7 operation. In the paper industry, uptime is a must. Line of business applications must be available to sales, production, shipping, receiving and management among many other departments.
The needs of Finch Paper were not unique in principle, but their needs were unique in regards to price versus solution. Their current environment contained a SAN infrastructure that provided the full allotment of feature sets, but commanded a heavy cost of maintenance. Their server room contained many racks of expensive physical servers that required expensive air conditioning units to cool the room. This combined with the amount of server racks created space constraints. The number of physical servers exceeded 50 and added a considerable amount of time to maintain. Hardware updates and maintenance consumed staff time and their focus was more on maintaining rather than innovating their environment. The number of physical servers was also creating backup/restore issues and was becoming harder to maintain. The sheer number of physical servers created wasted disk space that was being backed up unnecessarily.
The core network switch was being utilized as an access layer switch for desktops and not for core networking functionalities, as designed. These additional high port density cards were not being correctly utilized for server and core networking services. This was a wrinkle that needed to be ironed out to ensure the desired performance and reliability requirements of the delivered solution.
ADKtechs was tasked with designing a virtual environment to accommodate a high volume of traffic while maintaining consistent performance and high availability for the 1,000+ devices that would be utilizing these resources.
Prior to quoting the project, ADKtechs performed a thorough network assessment at the Finch Paper headquarters to properly size the VMware environment, ensuring it would be able to meet the needs of the customer. We also delivered a scope of services to guarantee the correct solution for the environment based off of our findings. The tools used in this assessment took into consideration the immediate needs of the server environment and future growth considerations.
ADKtechs put together a solution consisting of 6 host servers that contain four 6-core Intel processors accompanied by 416GB of RAM in each server. The total hardware resources being made available to the virtual environment was 1,344GB of RAM and 144 CPU Cores.
ADKtechs laid the groundwork for a robust, virtual environment with hardware that was needed to accommodate a solid, centralized storage solution that housed physical servers that would be converted to virtual servers. This would reduce the number of physical servers consuming electricity, generating heat and thus increasing cooling costs.
The centralized storage solution consisted of two Dell Equalogic SANs with redundant controllers, 70TB of RAW Tier 2 and Tier 3 storage. The SAN solution was proposed with only the functionality that the environment required and the ability to expand upon, if required. The SANs were configured to optimize where data was stored based on the performance requirements. If a database required faster read/write access, the data would be moved to a faster disk array to accommodate, as needed. This is commonly known as Compellent Data Progression.
Next, we needed to implement a local backup solution that could easily backup and restore their servers. Utilizing the technologies encompassed in the StorageCraft ShadowProtect product designed for virtual environments, we were able to perform backups of remaining legacy physical servers and virtual servers anywhere from daily to every 15 minutes on critical assets. Restore times in virtual or physical bare metal restores could easily be performed in minutes.
Lastly, we addressed the networking requirements by moving desktops from the core Nexus switch to stacked Cisco 3560s and setup 10GB Ethernet connectivity between the SANs and servers. We utilized 10GB ports on the stacked 3560s to create redundant 10GB network connections between server and SAN networks to handle the nearly 1,000 user/device environment.
The solution put forth by ADKtechs was designed, developed, and configured by two senior engineers on the ADKtechs staff. Together they have 30 years of technical experience and hold numerous Cisco security, VMware and Dell certifications. All physical servers were converted to virtual servers and turned off. Fifty physical servers now reside on a combined eight hardware devices, reducing the hardware to be maintained by 86%.
The SAN environment was setup in RAID 10 and RAID 50 configurations, providing an increase in disk performance of 33% while reducing Finch’s three-year cost of storage by $40,000 over the outgoing SAN environment. While we were able to reduce costs significantly and increase performance, we were also able to deliver a 24/7/365 4-hour response Maintenance Agreement on the SAN to ensure the high availability the client required. The SANs were configured to alert Dell of a predicted failure, automatically triggering the delivery of a replacement drive within 4 hours.
The proper utilization of the core networking assets added 10x the previously configured throughput and provided redundant network connections for both the SAN and server networks. The network cables were cleaned up utilizing cable management and assets were cabled for ease of troubleshooting. Server connections were cabled with red cables, SAN connections were cabled in black, uplinks were cabled in yellow, and so on. This clearly displayed the in-rack connections.
The project plan was created and executed over a six-week period of time. During this time frame, 50+ servers were virtualized, migrated, tested, and verified to be in 100% working condition. The only downtime experienced was encountered during scheduled outage windows. Server performance, client response times, and network performance was widely reported at the project’s end. Before wrapping up the project, extensive fail over tests were performed – all were 100% successful and took only minutes to perform full recoveries, should the virtual environment fail. As outlined by the project plan, we completed the project and received acceptance six-weeks to the day of project inception.
By project’s end, Finch Paper was able to spend less time maintaining their environment and more time innovating their organization. The resiliency of their systems increased while their costs plummeted. This project of virtualization and consolidation was delivered on time, on budget, and far exceeded their previous levels of performance while saving the organization more than $60,000 over other proposed technology paths.