Quick Take
Quick Take: Buying a 64Gb FC array does not buy 64Gb to the host, because the slowest hop sets the link speed. This checklist prices the whole path instead: HBAs, licensed switch ports, optics by reach, ISLs and roughly 25 percent spare ports, quoted as mirrored Fabric A and Fabric B.
Why Array-Only 64Gb Upgrades Fail
Quick Answer: Throughput follows the slowest hop. A 64Gb array port attached to a 32Gb HBA or switch runs at 32Gb. CapEx that upgrades only the array buys future-ready ports without host gain — and still needs dual-fabric hygiene that many RFQs forget.
Dell’s PowerStore Elite messaging (64Gb FC, 128Gb-ready) is pushing SAN refreshes into the same CapEx cycle as the array. Buyers who only re-quote the storage miss HBAs, MDS/Brocade ports, and optics — then discover negotiated 32Gb after cutover. Pair this checklist with PowerStore Elite BOM, PowerStore FC port planning, and 32Gb FC switch buying guide.
Shanghai TaoHan Industrial Co., Ltd. supplies FC HBAs, Cisco MDS / Brocade / Dell DS FC switches, optics, and PowerStore arrays with worldwide shipping — request a free SAN BOM risk assessment.
For the full multi-vendor framework, see the Enterprise SAN Storage Buying Guide 2026 (FC vs iSCSI vs NVMe-oF, AFA shortlist, HA, and five-year TCO).
| Failure pattern | What happens | BOM fix |
|---|---|---|
| Array FC only | Hosts stay 32Gb; no speed gain | HBA + switch lines on same RFQ |
| Single fabric quote | Maintenance window = outage | Fabric A and B as mirrored packages |
| Optics after install | Wrong reach / speed class | SFP matrix keyed to patch panel |
| No ISL plan | Islands when expanding | ISL ports + trunks sized upfront |
64Gb End-to-End vs Hybrid 32Gb
Quick Answer: Choose end-to-end 64Gb when critical hosts need the bandwidth and budget covers HBAs plus fabric ports. Choose hybrid when the array is 64Gb-ready but most hosts remain 32Gb for a phased year. Stay on 32Gb when KPIs are already met — do not buy 64Gb optics “just in case” without a written host upgrade plan.
Path A — End-to-end 64Gb
Array modules, switch ports, and HBAs on priority hosts all 64Gb-capable. Highest CapEx; cleanest ops story for Elite-era greenfields.
Path B — Hybrid negotiate-down
Array and core switches support 64Gb; existing 32Gb HBAs remain until refresh waves. Document which hosts are still 32Gb so capacity planning is honest.
Path C — Stay 32Gb deliberately
If IOPS/latency KPIs clear on 32Gb and port counts are the real constraint, expand 32Gb fabric first. Revisit 64Gb when hosts and array move together. See the 32Gb switch guide.
BOM Building Blocks
Quick Answer: Spec dual FC fabrics, switch OPNs and licensed port counts, HBAs per host (dual-port typical), optics by reach, ISL cables, and spare SFPs. Include array FC IO modules on the storage quote so SAN and array land as one project — not two calendars.
- Switches: Cisco MDS, Brocade, or Dell DS families with 64Gb-capable ports where Path A requires them; confirm NX-OS / FOS feature licenses.
- HBAs: 64Gb dual-port cards for priority hosts; retain 32Gb only under Path B with a written sunset list.
- Optics / DAC: Speed and distance matrix; label Fabric A vs B; 5–10% spares.
- Array side: PowerStore Elite (or other) FC modules counted in the storage BOM — see Elite guide and port planning.
- Services: Zoning, multipath validation, and cutover windows as named lines when in-house skills are thin.
Port Math and Dual-Fabric Rules
Quick Answer: Ports needed ≈ host initiator ports + array target ports + ISLs + growth (often ~25%). Deploy two independent fabrics. Each host should see each array port set down both fabrics — single-fabric “temporary” designs become permanent risk. Dual-port HBA → usually 2 initiators/host Do not steal host ports for ISL later For switch shortlists at 32Gb still.
| Component | Count hint | Notes |
|---|---|---|
| Host initiators | Hosts × ports per HBA | Dual-port HBA → usually 2 initiators/host |
| Array targets | Per appliance FC module plan | Align with PowerStore port guide |
| ISL / trunk | Per dual-fabric design | Do not steal host ports for ISL later |
| Spares | ~25% ports + optic kit | Growth and break-fix |
For switch shortlists at 32Gb still in estate, compare DS-7720B vs Cisco MDS before you standardize the next fabric buy.
Buyer Checklist Before You RFQ
Quick Answer: Attach speed strategy (64 end-to-end / hybrid / stay 32), host inventory with HBA generations, array FC module plan, dual-fabric switch sketch, and optics matrix. Ask for Fabric A and Fabric B as complete dated ATP packages alongside the array quote.
- Speed strategy written and signed by storage + server owners
- Host list with current HBA model and target upgrade set
- Array FC port count (Elite / T-series / other) frozen
- Dual-fabric switch OPNs and licensed port counts listed
- Optics length/speed matrix attached
- ISL and spare percentages included
- Zoning / multipath validation owner named
- Array and SAN on the same go-live calendar
Real-World Deployment Scenarios
Quick Answer: Map the three scenarios below to your workload, facility, and commercial constraints before freezing CapEx. Equal-BOM RFQs should cite which scenario drives the shortlist.
Scenario 1: Tier-0 Array Front-End Refresh (Performance Bound)
Workload Profile: All-flash arrays moving host ports from 32Gb to 64Gb FC / NVMe-oF.
Architectural Fit: Shortlist Gen7/Gen8 directors or fixed-port switches (G720/G730/DS-8820B class) with optics, PoD licenses, and ISL plans in one equal BOM.
Scenario 2: Dual-Fabric Bank Core (Availability Bound)
Workload Profile: Dual independent fabrics with hitless firmware expectations.
Architectural Fit: Migrate one fabric at a time; freeze TruFOS/optics policy before cutover so third-party optics do not block FOS upgrades mid-window.
Scenario 3: OEM Relabel Procurement (Cost Bound)
Workload Profile: Buyer can take Dell Connectrix / HPE / IBM OEM skins of Brocade silicon.
Architectural Fit: Cross-walk OEM MPNs to Brocade base before RFQ so LIST deltas are comparable—not three unrelated quotes.
Frequently Asked Questions
Quick Answer: 64Gb on the array alone does not deliver 64Gb to hosts. Mixed 32/64 links negotiate down. A SAN BOM needs switches, HBAs, optics, and dual fabrics. Upgrade the binding constraint first — ports or HBAs — never leave a single fabric as the production path.
Only if HBAs, switch ports, and optics also run 64Gb on that path. Otherwise you get negotiate-down links. Match the slowest hop to the KPI or upgrade end-to-end.
Usually yes via auto-negotiation at the lower speed. Fine for phased migration; do not claim 64Gb throughput on mixed hops.
Dual-fabric switches, HBAs, optics/cables, ISLs, spare SFPs, required fabric licenses, and array FC modules on the storage quote.
Upgrade the binding constraint. Port exhaustion → fabric first. Host bottleneck on a new 64Gb array → HBAs on critical servers first. Always keep dual fabrics.
Send host HBA inventory, array FC plan, and speed strategy via 提交 BOM / Contact TAOHAN or email info@021taohan.com. You receive a dual-fabric package with ATP signals from a global free-port hub.
Key Takeaway
Key Takeaway: Sign off one speed strategy with both the storage and the server owner before quoting: 64Gb end-to-end on priority hosts, hybrid with a dated 32Gb sunset list, or a deliberate 32Gb port expansion. Whichever you pick, buy both fabrics together with spare optics.