Enterprise IT Hardware in Connecticut

Yes — HPE ProLiant DL360 and DL380 Gen9 and Gen10, Dell PowerEdge R630, R730, R740 and R750, Lenovo ThinkSystem and Supermicro, as complete systems and as parts for specific models.

Yes. Tell us the model and generation — DL380 Gen9, R730, ThinkSystem SR650 — and we quote drives, caddies, memory, processors, controllers and power supplies against that platform.

Enterprise SSDs with power loss protection and an endurance rating matched to the workload — typically mixed use at around 3 DWPD. Consumer flash under sustained transactional writes wears out far faster than expected.

Capacitors that let an SSD complete in-flight writes when power is lost. It is the difference that matters most between consumer and enterprise flash, and it matters most in transactional systems.

High-capacity CMR nearline drives in a double-parity array, with LTO tape behind them for long retention. Archives grow continuously and are read far more than written.

Yes — LTO-6, LTO-7, LTO-8 and LTO-9 drives and media, plus autoloaders and libraries from HPE StoreEver, IBM, Quantum and Overland.

LTO drives generally read two generations back and write one generation back, with limits that changed from LTO-8 onward. Tell us the drive and the media you hold and we will confirm before you buy either.

600GB 15K SAS, 900GB SAS, 1.2TB, 1.8TB and 2.4TB 10K SAS, and 4TB, 8TB, 10TB, 12TB, 16TB and 20TB nearline drives, in 12Gb/s SAS and 6Gb/s SATA.

Seagate Exos, Western Digital Ultrastar and Toshiba, plus vendor-coded HPE, HP, Dell, IBM, Lenovo and NetApp stock.

Spindle speed trades capacity for latency. 15K SAS suits transactional workloads, 10K is the general-purpose middle, and 7,200 RPM nearline gives the most capacity per bay for archives.

A 7,200 RPM high-capacity drive with a SAS interface rather than SATA. You get capacity economics with SAS queueing and dual-port capability, which is why arrays favour it.

1.92TB, 3.84TB and 7.68TB in 12Gb/s SAS, 6Gb/s SATA and NVMe U.2, plus M.2 — Samsung, Micron, Intel, Seagate Nytro, Kioxia and vendor-coded HPE and Dell.

Endurance. Read intensive drives are rated around 1 DWPD and suit mostly-read workloads; mixed use is typically 3 DWPD for databases and virtualisation. A read intensive drive in a write-heavy role wears out early.

Drive writes per day and terabytes written — how much can be written over the warranted life. Match the rating to the workload rather than assuming any enterprise SSD suits any role.

Both are 2.5 inch NVMe form factors. U.3 backplanes accept U.2 drives, but U.2 backplanes do not accept U.3 — the compatibility runs one way only.

512n, 512e and 4Kn describe how a drive presents its sectors. Mixing formats in one array is refused or unpredictable, and it frequently surfaces during a rebuild rather than at installation.

Yes, always. SMR write performance collapses part-way through sustained writes, and a rebuild onto an SMR drive can fail outright. We state CMR or SMR on every listing.

Yes. The drive reports it exactly and we check before listing. At volume we can state the range across the batch.

Yes — Dell PERC H730, H730P, H740 and H965i, HPE Smart Array P440ar and P408i, plus LSI and Broadcom HBAs and QLogic and Emulex fibre channel HBAs.

Yes. This is the part that fails silently — the controller drops the array into write-through and performance falls off with no obvious error. A replacement pack is cheap relative to the symptom.

A host bus adapter that passes drives through to the operating system without RAID abstraction — an LSI 9207-8i or 9300-8i, for instance. Software-defined storage and ZFS need it.

Yes — Dell drive caddies and HPE Smart Carrier trays, plus drive blanks. We stock them alongside the drives, because a drive without the right caddy will not go in the bay.

No. Smart Carrier trays changed across generations and do not cross. Tell us the server model and generation and we supply the right one.

Yes. Rails are chassis and rack specific, and a server without the right rails does not mount safely. Tell us the model and rack depth.

Yes. Risers determine which PCIe slots are usable and at what lane width, which is why an adapter that fits physically may still run slower than expected.

Yes. Heatsinks are frequently processor and chassis specific, and a CPU upgrade sometimes needs a different heatsink.

16GB, 32GB, 64GB, 128GB and 256GB DDR4 RDIMM and LRDIMM registered ECC in DDR4-2400 PC4-19200, DDR4-2666 PC4-21300 and DDR4-3200 PC4-25600, plus DDR5-4800 and DDR5-5600 and DDR3-1600 PC3-12800.

Samsung, Micron, SK Hynix, Kingston and Crucial, plus HPE and Dell coded modules where the platform expects them.

RDIMM buffers the address and command lines, LRDIMM buffers the data lines too and allows much higher total capacity, and UDIMM is unbuffered. They do not mix in the same system.

Population rules. Filling more ranks or more slots per channel forces the bank to downclock, and the processor also has a maximum supported speed. The modules are not faulty.

You can, but everything runs at the slowest common speed and unbalanced population costs you channel performance. Populate in matched sets per the platform’s memory guide.

Yes, along with Xeon Scalable Bronze, Silver, Gold and Platinum from 8 to 28 cores, and AMD EPYC and Opteron.

No. Both sockets need matching models. A mismatched pair either refuses to boot or runs on one socket.

Frequently yes, and the BIOS has to be updated before the new processor goes in. Tell us the server and current firmware level and we will flag it.

Yes, by wattage — 495W, 750W, 800W, 1100W and above — for HPE, Dell and Supermicro platforms.

Generally no. Redundant supplies should match on wattage and usually on part number; a mismatched pair commonly logs a fault and may not provide redundancy at all.

Yes, including UPS battery replacements. Batteries degrade continuously and a self-test draws far less than a real outage, so test under real load periodically.

Yes, including SFP-10G-SR, QSFP+ and QSFP28.

Often not. Many switches only accept transceivers coded for that vendor and simply refuse others. Tell us the switch make and model and we supply the correct coding.

Yes — Intel X520 and X710, Mellanox ConnectX and Broadcom adapters in PCIe, OCP and mezzanine form factors, plus Cisco Catalyst and comparable PoE+ and 10GbE switches by port count.

Not if it has come off a cold truck into a warm room. Condensation can form internally. Allow several hours to reach room temperature first.

If the packaging still feels cold to the touch, it is not ready. Several hours for a server, less for a drive or scanner.

Heated buildings have dry air, and dry air holds charge. Memory, controllers and bare drive electronics are what gets damaged.

The component works, passes testing and fails earlier than it should. That is most static damage, which is why handling practice matters even when nothing appears to go wrong.

Touch the chassis before touching components. It is free, takes a second, and prevents most incidents. Re-ground each time you return to the equipment.

Not against the component — it generates static. Use an anti-static bag first with padding outside it.

Yes. Zebra and SATO industrial printers suited to part marking, traceability and maintenance record labelling, with printheads and spares.

Labels have to remain readable through a component’s service life, which usually means thermal transfer with a resin ribbon rather than direct thermal. Tell us the surface and environment.

Yes. Zebra and SATO printers for patient identification, wristband and specimen labelling, plus genuine replacement printheads.

Tell us the temperature range, the surface and the required life. Adhesives are rated for both the application and storage temperature, and those are frequently different.

Small labels with dense data usually need 300dpi rather than 203dpi. Tell us the label size and what has to fit on it and we will advise.

Read the part number off the existing printhead or the printer label. Printheads differ by resolution for the same printer model and do not cross between models.

A dead heating element in the printhead. It appears in the same position on every label, does not heal, and more elements will follow.

Calibration. The printer detects label boundaries with a sensor, and changing media size, liner or gap requires recalibrating.

No. Data Matrix and QR codes are 2D and require an imager. A 1D laser cannot read them at all.

Yes, and it is specific to the interface. We supply the correct cable with the scanner rather than selling it separately.

Yes, to every address in the state — Fairfield County, the Hartford area, New Haven, the shoreline and the eastern counties.

Connecticut is on the eastern seaboard corridor from West Palm Beach and standard ground reaches the whole state well.

No. We ship from West Palm Beach, Florida, and we would rather state that plainly than imply a distribution network we do not have.

Yes. Buildings with loading dock restrictions, delivery windows or no street-level access can add a day regardless of service, so tell us about access constraints when you order.

Free shipping applies to qualifying orders. Heavier and expedited shipments are quoted separately.

Yes. We accept purchase orders from organisations, offer Net-30 on approval, and complete vendor registration forms and W-9s as required.

Yes. Purchase orders, formal quotations and vendor registration are all standard for us.

Yes. Send the list with part numbers, quantities and the target system for each line, and you get it back priced per line with exact part number, condition and availability stated.

Yes — servers, drives, caddies, memory, processors, controllers, power supplies, transceivers and tape quoted together against the platforms you run.

We state a validity period on every quote. Ask for longer if your approval process needs it and we will tell you what we can hold.

Sales tax is applied where we are required to collect it. If your organisation is tax exempt, send your exemption certificate before ordering.

Our target is one business hour, Monday to Friday, 9:00 AM to 6:00 PM Eastern — the same time zone as you.

We state the condition on every line. Ask for the specific condition, the testing performed and the drive hours where applicable.

Testing is performed where applicable and we will tell you what was done. A power-on check and a full surface test are different things and we do not describe one as the other.

30 days. Raise a problem promptly — the sooner it is raised, the more options there are.

Yes, after it has acclimatised and before it goes into production. Failures cluster early in life and transit damage frequently does not show until equipment runs under load.

Tell us the platforms and printer models you run and your refresh timing. It commits you to nothing and it changes what we can do when something fails.

Yes. Several of our guides recommend against a purchase where the workload does not justify it, and we apply the same approach to quotes.

Contact us with a list of what is coming out and we will tell you what it is worth and what is worth keeping as spares.

Call (954) 601-5103 or email sarah.jane@techsellerusa.com. Business hours are Monday to Friday, 9:00 AM to 6:00 PM Eastern.