Optimizing Storage with a New Server
Seeking for Insights on Setup
As my business grows, I’ve realized the need to expand my cloud infrastructure. Initially, the plan was to replace my original server (the "First Server") with a new, more powerful system (the "Second Server"). However, after much thought, I’ve decided to keep both servers to improve uptime, expand capacity, and increase flexibility. This article outlines the final plan for the deployment of both servers. I’d love your thoughts as I prepare to move forward.
First Server: Current Setup
The First Server was my first step into the cloud, serving as the foundation of my business operations. Its setup prioritizes redundancy and performance, ensuring smooth handling of critical workloads. While it has served me well, it’s time to optimize and repurpose it to complement the Second Server.
This server is equipped with 96GB of DDR3 RAM and a storage configuration focused on reliability. The primary storage consists of three 4TB drives in a 3-way mirror, providing excellent redundancy. Additionally, two 1TB NVMe drives are partitioned into a 500GB mirrored storage tier for fast performance and a 500GB read cache (L2ARC) to accelerate slower drives. A single 8TB drive is reserved for disaster recovery, supplemented by cloud backups for extra safety.
Hardware:
- 1RU server with 96GB DDR3 RAM.
Storage:
- 3x 4TB drives in a 3-way mirror, ensuring data redundancy.
- 2x 1TB NVMe drives partitioned as:
- 500GB in RAID 1 (mirror) for fast storage.
- 500GB for L2ARC (cache) to boost read speeds.
- 1x 8TB drive for disaster recovery, with supplementary cloud backups.
Planned Changes to the First Server
Rather than replacing the First Server, I plan to reconfigure it for less intensive workloads, freeing up high-performance components for the Second Server. The 3-way mirror of 4TB drives will be replaced with a RAID 10 setup using four 4TB drives. This will provide improved performance while maintaining redundancy. The NVMe drives, which are critical for fast operations, will be moved to the Second Server. Additionally, the 8TB backup drive will be repurposed for other uses.
These changes ensure the First Server remains a vital part of the infrastructure while adapting to its new role as a support system for the Second Server.
- Replace 3x 4TB drives in a 3-way mirror with 4x 4TB drives in RAID 10 (striped 2-way mirrors).
- Move the 2x 1TB NVMe drives to the Second Server for performance-critical tasks.
- Repurpose the 1x 8TB backup drive for another system.
Introducing the Second Server
The Second Server is my latest addition to the cloud infrastructure. It is designed to handle high-demand tasks like databases, backups, and web hosting while providing room for future growth. This server combines powerful hardware with repurposed components from the First Server, offering both performance and cost efficiency.
Equipped with 256GB of DDR4 RAM and eight drive bays, the Second Server is built for scalability and speed. The storage configuration includes six 1TB SAS HDDs, two 500GB SATA SSDs and the two 1TB NVMe drives migrated from the First Server. This setup enables it to become the backbone of my operations, handling critical workloads with ease.
- Hardware:
- 1RU server with 256GB DDR4 RAM.
- Storage:
- 6x 1TB SAS HDDs for capacity and redundancy.
- 2x 500GB SATA SSDs for hot spares and caching.
- 2x 1TB NVMe drives, migrated from the First Server for high-speed operations.
Connecting the Servers: High-Speed Backhaul
To ensure seamless communication between the First and Second Servers, I’m introducing a 10Gbps backhaul connection. This high-speed link will enable efficient data transfers, faster synchronization of backups, and smooth virtual machine migrations. It is a critical feature for integrating the two servers into a unified system.
The backhaul will use SFP+ ports and is designed to scale up to 20Gbps in the future through link aggregation. This ensures the infrastructure remains future-proof while supporting the growing demands of my business.
- 10Gbps SFP+ link, expandable to 20Gbps via link aggregation.
- Enables faster:
- Backup synchronization.
- Database replication.
- Virtual machine migrations.
Storage Configuration Plan
First Server:
Mostly used for backup and redundancy, or services where performance is not critical.
- Storage:
- 4x 4TB drives in RAID 10 for a balance of storage space, redundancy and easy to upgrade if needed.
Second Server:
Used for performance critical operations like web hosting and VPS of clients.
SAS HDDs:
Partitioning of 6x SAS HDD 1TB:
- 100GB for the OS in a 4-way mirror, with two hot spares.
- 900GB for data in 2x 2-way mirrors stripped, with two hot spares.
- ARC Cache:
- 8GB for the OS.
- 128GB for data.
NVMe and SSDs:
- Fast Storage Tier:
- 2x 500GB NVMe drives in RAID 1 (mirror) for performance-critical workloads.
- 2x 500GB SATA SSDs as hot spares for reliability of Fast storage used for databases.
- Caching Layer:
- Remaining NVMe capacity used for L2ARC (read cache) for the data on the SAS HDD in case the ARC is not enough.
Why This Setup Works
This setup offers a balance of performance, redundancy, and scalability while reusing existing hardware to keep costs under control. By retaining both servers, I can improve uptime, expand my capacity, and adapt to future growth.
As time goes and the storage starts to degradete I can easily upgrade the SAS HDD 6Gbps for SAS HDD 12Gbps and the solid storage to enterprise grade solid storage.
- Improved Uptime: Keeping both servers reduces downtime and ensures reliability.
- Enhanced Performance: NVMe storage and caching layers speed up critical tasks.
- Redundancy: RAID configurations and hot spares minimize the risk of data loss without imediate human intervention.
- Scalability: The 10Gbps backhaul allows for seamless expansion and resource sharing and can be expanded to 20Gbps.
- Cost-Effective: Reusing hardware maximizes value while minimizing expenses.
Your Feedback
This plan reflects months of thought and preparation, but I’m always open to suggestions. Have I overlooked anything? Are there areas where this setup could be improved? Your advice is invaluable as I finalize this deployment. Thank you for taking the time to help!
