What is the best 2000$ Hadoop/Big Data Software Development Home Workstation Configuration?

[thennarasan]

I am planning to configure a good computer which will help me learn Apache Hadoop, HBase, Apache Spark, Apache Hive and Big Data, Cloud Computing and Machine Learning related technologies.
I found many benchmarkings for gaming and content creation professionals but not much for software development professions that too Big Data Workstation configuration despite its popularity 🙁
I am currently taking Big Data Specialization MOOC in Coursera and planning to take Data Science MOOCs in Udacity.
Currently I have a 8.5 year old Core2Duo 2GB RAM PC, hence planning to assemble a new Work Station computer to take advantage of the latest CPU, DDR4 RAM and SSDs available.

Location: India. My budget is 2000$. 1,25,000 INR

My Requirements:

- Big Data / Web Development / Machine Learning ( 6 hours max in weekdays, 24 hours in the weekends 😀 )

- Will be running Linux VM

- Cloudera Sandbox VM

- Hortonworks Sandbox

- Extreme Web Browsing in Chrome and Firefox

- No Games, No Overclocking

Will be using my existing monitor, wireless keyboard and mouse.

I have selected the following configurations.

#1. intel i7 - 6 Core DDR4 Config - Costly but Expandable RAM Slots in the future:

Intel Core i7 5820k 6 Core 15 MB L3 Cache
Corsair 32 GB 2400MHz DDR4 RAM
500 GB Samsung 850 EVO SSD
4 X 1 TB 7200 rpm Seagate HDD
NVIDIA Quadro K620 2GB
Windows 10 Pro
Motherboard : ASUS x99 Sabertooth or Deluxe or WS or Pro ?? (Not decided, confused still)

#2. Intel i7 - 4 Core DDR4 - Config - Similar to the above config but latest Skylake CPU only 4 DIMM slots

Intel Core i7 6700k Core 8 MB L3 Cache
Corsair 32 GB 2400MHz DDR4 RAM
500 GB Samsung 850 EVO SSD
4 X 1 TB 7200 rpm Seagate HDD
NVIDIA Quadro K620 2GB
Windows 10 Pro

#3. intel i7 - 4 Core DDR3 Config - Economical but not expandable.

Intel Core i7 4790K 4 Core 8 MB L3 Cache
Corsair 32 GB 2400MHz DDR3 RAM
500 GB Samsung 850 EVO SSD
4 X 1 TB 7200 rpm Seagate HDD
NVIDIA Quadro K620 2GB
Windows 10 Pro

#4. intel Xeon E5 1620 v3 - ECC DDR4 - Too costly but Server grade configuration.

Intel Xeon E5 1620 v3 10 MB L3 Cache
Samsung 16GB DDR4 ECC 2133 MHz RAM
HP Z Turbo Drive G2 256 GB (PCIe SSD)
2 X 1 TB 7200 rpm WD HDD
NVIDIA Quadro K620 2GB
Windows 10 Pro

Kindly suggest me a good configuration from the above configurations or your own suggestion are welcome as well.

Thanks in Advance!

 

[cst1992]

I'd say you should go with the first one. More CPU cores should help.
A GTX card will be cheaper than Quadro, especially here in India.

I don't think you need Quadro unless you're professionally doing rendering(The high-end rendering cards are used to render large models such as a Boeing 777 jet down to the last component).

As long as the card supports CUDA, you should be fine. GTX cards do.

And I'd say invest more in the card. A 13,700 Quadro card won't be very powerful, plus way more expensive than a geforce card(twice the price per gigaflop).

 

[thennarasan]

I'd say you should go with the first one. More CPU cores should help.
A GTX card will be cheaper than Quadro, especially here in India.

I don't think you need Quadro unless you're professionally doing rendering(The high-end rendering cards are used to render large models such as a Boeing 777 jet down to the last component).

As long as the card supports CUDA, you should be fine. GTX cards do.

And I'd say invest more in the card. A 13,700 Quadro card won't be very powerful, plus way more expensive than a geforce card(twice the price per gigaflop).

Hi cst1992,

Thanks for the response!

I will go with the first configuration with i7 5820k 6 core processor.
For the graphics card part, mostly I will be doing Data Visualization and may watch 3D movies if possible.
Will not play games. I have checked online Quadro is the one suggested for workstation level configurations and GTX cards are for gaming PCs. So can you suggest me some decent graphic card for the aforementioned need within 10000 INR. I checked the price of NVIDIA Quadro K620 in a computer retailer shop, price is 10500 INR.
So what do you suggest?
Thanks!

 

[cst1992]

Workstation, sure, but there's a variety of workloads one can consider. As I said, Quadro cards are for workstation rendering tasks. A K620 is only as powerful as a GT 740, a card which is only around 70% of the price(or lower). Get a Geforce card. I am not talking about gaming, I'm talking about CUDA(which is not gaming).

I have attended a workshop in college where they taught high-performance computing, and they were not using Quadros.

That said, a Geforce card that you'll be able to get is the 750Ti, with 2 GB of VRAM.

 

[thennarasan]

Workstation, sure, but there's a variety of workloads one can consider. As I said, Quadro cards are for workstation rendering tasks. A K620 is only as powerful as a GT 740, a card which is only around 70% of the price(or lower). Get a Geforce card. I am not talking about gaming, I'm talking about CUDA(which is not gaming).

I have attended a workshop in college where they taught high-performance computing, and they were not using Quadros.

That said, a Geforce card that you'll be able to get is the 750Ti, with 2 GB of VRAM.

Hi Cst1992,

I was unaware of CUDA technology altogether. Glad you mentioned about that.

Within 10k Geforece 750Ti is really good. As the no of CUDA cores are good for the future, so I found Geforce 960 (around 200$ ) with 1024 over 750 Ti with 640 CUDA cores.

As I am planning to go for Asus x99 motherboard so thought of selecting graphics card from Asus as well.

Do you think Aus Gefore GTX 960 a good option?

Specification of Geforce GTX 960 can be found here.

Thanks,
Thennarasan

 

[cst1992]

A general practice is to get the fastest card that you can afford.
There is no advantage when using hardware from the same vendor, apart from the fact that you can download drivers from a single website. I download drivers straight from the official NVIDIA website(geforce.com) myself.

The Strix 960 is a good card. It's faster than the 750Ti, and worth the money.

I use techpowerup for looking up specs. I find it has more exhaustive information than the NVIDIA page. Just google "<card> techpowerup" and select the link which says " <card> | TechPowerUp GPU Database".

Here's the 960's link for you: ASUS STRIX GTX 960 DirectCU II OC | techPowerUp GPU Database

 

[thennarasan]

A general practice is to get the fastest card that you can afford.
There is no advantage when using hardware from the same vendor, apart from the fact that you can download drivers from a single website. I download drivers straight from the official NVIDIA website(geforce.com) myself.

The Strix 960 is a good card. It's faster than the 750Ti, and worth the money.

I use techpowerup for looking up specs. I find it has more exhaustive information than the NVIDIA page. Just google "<card> techpowerup" and select the link which says " <card> | TechPowerUp GPU Database".

Here's the 960's link for you: ASUS STRIX GTX 960 DirectCU II OC | techPowerUp GPU Database

Hi Cst1992,

Thanks buddy!

I will go with Strix 960. One final clarification, that is none other than the Power Supply !

My New System configuration is as follows: What power supply and Cooler should I go for?

Processor : Intel Core i7 5820k
Motherboard : Asus X99-A or X99-Pro or X99-Deluxe (May be anyone)
RAM: (4 X 8 ) 32 GB Corsair DDR4 2400 MHz
SSD: 256 GB Samsung 840 Evo
HDD: 2 X 1 TB Seagate 7200 rpm
Graphics Card: Asus NVIDIA STRIX-GTX960-DC2OC-2GD5
OS : Windows 10 Pro
Cabinet : Not Sure yet, Will decide after seeing some in the shop. But will choose one from Cooler Master or Corsair.

Thanks,
Thennarasan

 

[cst1992]

I'd say get the X99-A. It's unlikely you'll need a more powerful motherboard, not to mention there's a significant cost difference in the three models.
Also, a single 2TB hard drive will be cheaper(6000) instead of two 1TB(2x3600). Unless you're setting up RAID to take advantage of more performance, you should get a single 2TB drive.

So your build comes to this:

PCPartPicker part list

CPU: Intel Core i7-5820K 3.3GHz 6-Core Processor
Motherboard: Asus X99-A ATX LGA2011-3 Motherboard
Memory: Corsair Vengeance LPX 32GB (4 x 8GB) DDR4-2400 Memory
Storage: Samsung 850 EVO-Series 250GB 2.5" Solid State Drive
Storage: Seagate Barracuda 2TB 3.5" 7200RPM Internal Hard Drive
Video Card: Asus GeForce GTX 960 4GB Video Card
Operating System: Microsoft Windows 10 Pro OEM (64-bit)

I added these parts(you might end up with different ones; I added these for the sake of posting a complete build):

CPU Cooler: Cooler Master Hyper 212 EVO 82.9 CFM Sleeve Bearing CPU Cooler
Case: Corsair 400R ATX Mid Tower Case
Power Supply: Antec 750W 80+ Gold Certified Fully-Modular ATX Power Supply

The power supply is keeping in mind that you may want to add another card in the future, so you will not need to upgrade the power supply. AFAIK your motherboard supports three graphics cards installed at once.
The CPU cooler and case are random popular ones you might want to get; but you could go with different ones too. If you go in the store and have any problems, be sure to post back. And try online too: you might get some killer deals.

 

[thennarasan]

I'd say get the X99-A. It's unlikely you'll need a more powerful motherboard, not to mention there's a significant cost difference in the three models.
Also, a single 2TB hard drive will be cheaper(6000) instead of two 1TB(2x3600). Unless you're setting up RAID to take advantage of more performance, you should get a single 2TB drive.

So your build comes to this:

PCPartPicker part list

CPU: Intel Core i7-5820K 3.3GHz 6-Core Processor
Motherboard: Asus X99-A ATX LGA2011-3 Motherboard
Memory: Corsair Vengeance LPX 32GB (4 x 8GB) DDR4-2400 Memory
Storage: Samsung 850 EVO-Series 250GB 2.5" Solid State Drive
Storage: Seagate Barracuda 2TB 3.5" 7200RPM Internal Hard Drive
Video Card: Asus GeForce GTX 960 4GB Video Card
Operating System: Microsoft Windows 10 Pro OEM (64-bit)

I added these parts(you might end up with different ones; I added these for the sake of posting a complete build):

CPU Cooler: Cooler Master Hyper 212 EVO 82.9 CFM Sleeve Bearing CPU Cooler
Case: Corsair 400R ATX Mid Tower Case
Power Supply: Antec 750W 80+ Gold Certified Fully-Modular ATX Power Supply

The power supply is keeping in mind that you may want to add another card in the future, so you will not need to upgrade the power supply. AFAIK your motherboard supports three graphics cards installed at once.
The CPU cooler and case are random popular ones you might want to get; but you could go with different ones too. If you go in the store and have any problems, be sure to post back. And try online too: you might get some killer deals.

Really astonished by your helpful feedback ! Thanks a lot Buddy

I heard or seen somewhere online 2TB HDDs are not supported by Windows 8.1. That's why chose to go for 2 X 1TB.

You are right! X99-A makes more sense to me while comparing other options!

Thennarasan

 

[thennarasan]

I purchased the below configuration for $2300 ~ 1,51,000 INR

CPU: Intel Core i7-5820K 3.3GHz 6-Core Processor
Motherboard: Asus X99-PRO ATX LGA2011-3 Motherboard
Memory: Corsair Vengeance LPX 32GB (4 x 8GB) DDR4-2400 Memory
Storage: Samsung 850 EVO-Series 500GB 2.5" Solid State Drive
Storage: Seagate Barracuda 2TB 3.5" 7200RPM Internal Hard Drive
Video Card: Asus GeForce GTX 960 4GB Video Card
Operating System: Microsoft Windows 10 Pro OEM (64-bit)
CPU Cooler: Cooler Master Hyper 212X 82.9 CFM CPU Cooler
Case: Cooler Master MasterCase Pro 5 ATX Mid Tower Case
Power Supply: Antec 750W 80+ Gold Certified Fully-Modular ATX Power Supply

I am thankful for your instant feedbacks cst1992

Thennarasan

 

[cst1992]

Great config. Have fun!