Moving from SoftLayer to HiVelocity or...

Portal Home > Knowledgebase > Articles Database > Moving from SoftLayer to HiVelocity or...

Posted by uninet, 11-21-2016, 06:23 PM
Hi, everyone, I posted about a month ago that I was thinking about moving from SoftLayer. I confirmed with them that moving to a new system -- even a change as modest as going from my E3-1270v2 to a like-configured v5 -- would boost my hosting bill from $189/month to $430+. Looking at HiVelocity's current Black Friday server sale, I'm thinking it is time to take the plunge. I'm debating the exact server, though: should I do a E3-1270v5 (faster cores) or a D-1540 (more cores)? I'm leaning towards their preconfigured D-1540 (32 GB of RAM) with a 480GB SSD and 1 TB SATA drive. (I like having a drive to back up to on the server and the E3-1270v5 preconfigured systems don't have that option). Assuming I prepay annually, I'll save approximately $54/month. My questions: can anyone speak to the experience of being a HiVelocity customer after having used SoftLayer? And, any thoughts on those two processors in comparison? If someone thinks I should look at another provider that has a similar configuration, I'm happy to hear about that, too, but this looks like the best deal I've seen from a seemingly reputable provider. Thanks, Tim
Posted by SolaDrive - John, 11-22-2016, 12:03 AM
What are you using the server for? If its for web use I would go with the E3-1270v5 as its going to give the best performance for apache/mysql requests with having a higher clock rate.
Posted by Mike_A, 11-22-2016, 12:53 AM
The E3 would probably be better in most use cases, HiVelocity is definitely a good choice. I'm sure you can contact them to get an extra 1TB drive put in for a small fee even if it is a pre-configured server.
Posted by baremetalservers, 11-22-2016, 03:44 AM
Hi, What exactly server configuration is needed? CPU, RAM, HDD, RAID array, IP KVM, Network speed and bandwidth? I would recommend you to choose the E3 server v3 or v5 version. Also, could you remind us what type of project do you have? Thank you for your answer in advance. If we have more details, we could recommend you the right solutions.
Posted by NortheBridge, 11-22-2016, 05:45 AM
While I can't speak about HiVelocity, although I have heard good things about them, the one thing SoftLayer has over everyone else is their network is very well connected and you have a lot of capabilities that you won't find at other providers. With that said, it seems like IBM is trying to kill the SoftLayer brand; the prices of servers used to be fair market value whereas now you are paying about as much as it would cost to actually buy the server within 2-3 months of service start. When we first saw the trends of SoftLayer's pricing moving upwards we assumed it's to align with IBM's enterprise focus but not it seems like it's just to push people towards the already failed Bluemix platform. Unfortunately, no one else quite compares to SoftLayer - you will sacrifice something no matter what but when it becomes prohibitive (such as, for example, $400+ for an E3-1270v5 server) you just have to bite the bullet and find someone else. Something tells me that within the next few years, if that, SoftLayer won't actually exist anymore.
Posted by SreePriya, 11-22-2016, 06:00 AM
I am not having any personal expereince with Hivelocity but have many clients who is using Softlayer. The support from Softlayer is commendable. and never had any bad expereince from them. Is the rate only your issue or are you having any other issues with softlayer ?
Posted by Hacksley, 11-22-2016, 07:22 AM
If the webmaster has decided to move his website to the other provider, the reasons must be rather serious. If the problems are light, they are solved fast and easily.
Posted by Srv24x7, 11-22-2016, 09:39 AM
Hi, Never went into any issue on any client with HiVelocity. It is good one. I don't think there would be any issues of disatisfaction you will get with this move.
Posted by TmzHosting, 11-22-2016, 12:25 PM
We have worked with Hivelocity, and have had no issues. Remember Softlayer is on a completely different level than they are, they operate on a much bigger scale. - Daniel
Posted by madRoosterTony, 11-22-2016, 12:26 PM
Based on your last post, you would be much better with the E3-1270v5 CPU over the D-1540. The D-1540 is good for someone that has a light weight web application, but a ton of traffic. The issue with the D-1540 is the core speed is low, any web application that puts the core under load is much slower to respond. Therefor slowing down the overall response of the server As others have mentioned, you should be able to add an additional drive to any of HiVelocity's servers. Last edited by madRoosterTony; 11-22-2016 at 12:29 PM.
Posted by Lakjin, 11-22-2016, 06:28 PM
No experience with Softlayer but we are with HiVelocity and they are great. Highly recommended.
Posted by uninet, 11-22-2016, 10:09 PM
As Tony noted, I have a web server. Actually, updating on that, my current server (1270v2 with 16GB of RAM, a RAID1 of HDDs, 1Gbps port, etc., running on RHEL6) is now handling the load just fine. Nonetheless, as Northebridge noted, it seems like I'll need to migrate at some point since SL has pretty much gone off the rails on pricing and, if I can get a newer server and lower the monthly bill, that's hard to complain about. It doesn't look like HiVelocity is very willing to do any customization to their preconfigured servers, though, so if I go that route, I either need to live with just one SSD (and maybe get their 24x7 backup service?) or I'd have to get the D-1540. I'm a little antsy about having the server contain only one drive, although maybe I shouldn't be, assuming everything is regularly backed up...
Posted by hivelocitygm, 11-22-2016, 10:29 PM
We offer both an instantly deployed and a fully customizable version of the E3-1270 v5 server. Instant Deploy servers are already built so hardware changes are not a available. Obviously the customizable version can be built however you want it.
Posted by netstepinc, 11-23-2016, 12:13 AM
I'm currently on a dual SSD with RAID-1 with ASmallOrange.com, a massive cost savings after 10 years with a dedicated Rackspace server. However, I'm considering moving to a single SSD in HiVelosity too. I know all hardware can fail but SSDs are extremely reliable. My fixation with redundant drives dates back to IDE and SATA drives so I'm trying to step back and reconsider that requirement. I signed up with ASO a couple years ago on a 50% off deal and prepaid 2 years. However, now I either need to pay 30% more or strip the server down to save some money...and right now I need to save some money due to external factors. The HiVelocity deals are looking pretty awesome, especially with the annual payment discount. The configurable servers are a little more than their pre-canned ones which makes sense because they've got the cost savings of an assembly-line production. My theory is to get one of their instant servers, and use their cheap cloud storage for more frequent database backups for eCommerce site databases thereby reducing the potential data loss to maybe an hour instead of a whole day. The trade off is saving quite a bit of money while minimizing the data loss risk. ...this is a working theory...I've been leasing servers since 2002, but I'm no expert. One thing I can say about ASO is the server has run extremely well, and I've had few issues. Their support is painfully slow and chat support is typically routed to India to someone with very little power to solve any problems. Some of my tickets have gone 3-8 hours with no response, which leaves a pretty sour taste in my mouth, even with begging the chat person to get someone's attention. In contrast, HiVelocity's 24/7 support is at their data center with people who can actually help solve problems via ticket > chat and phone if absolutely necessary. That's a BIG plus in my book.
Posted by net, 11-23-2016, 02:07 AM
I would go for Software Raid1 instead of 1 drive. No matter what kind of drive it will be. It is better to be safe than sorry. Of course, remote backup should always be in the picture.
Posted by netstepinc, 11-23-2016, 02:45 AM
Yes... That was plan A... Thank you for the reality check. I'm in a bit of a fiscal pickle so I'm trying to save a few bucks. Of course if the drive fails and my customers come unglued then I'd feel pretty stupid after having soft RAID for the last 12 years on my other servers.
Posted by instantlyloved, 11-23-2016, 10:06 AM
Well, if hivelocity mentioned they are providing full customization of servers suitable for your needs why not to give them a try? They seems decent
Posted by uninet, 11-23-2016, 01:53 PM
Here's a question: is there some way to do a RAID-like setup between an SSD and an HDD, where the SSD would essentially serve as a cache, but would write a duplicate to the HDD when the load permitted it? I like the idea of moving to an SSD-base server, but it seems like this sort of setup would be the ideal arrangement -- performance for the front facing server, but redundancy if something went wrong... I seem to recall that one could do something like that with ZFS, but it has been a few years since I looked into it.
Posted by madRoosterTony, 11-24-2016, 02:35 AM
I know for sure with ZFS you can do RAID1 with two different partitions, so the ~500GB SSD as full partition and say 1TB drive in two partitions as 1 - 500GB partition to RAID1, 2nd partition to Backup. The problem with this is you only get the performance of the slowest drive as the system has to wait until both drives are written too. So having the SSD becomes usless basically. I do not know anyone that has tried it with MDADM (Linux Software RAID), but the same theory on speed should be expected. The idea of SSD Cache is not to mirror the data, but move the most referenced data to a small SSD compared to a larger HDD that holds everything. Its not that it gives you redundancy, but instead it gives you performance boost under some situations.
Posted by madRoosterTony, 11-24-2016, 02:39 AM
Also should be noted and I have made it publicly known, I am against using any type of RAID for redundancy, as I have seen many, MANY, times where a client will use RAID and then lapse on their backups thinking they are safe. However, what happens the majority of the time is, instead of a hard drive directly failing, there will be some type of minor failure that will then cause corrupted data to be written across all drives in the RAID system, therefor making the redundancy of RAID being a non valuable thing. RAID should be used for performance and space increase only. No one should ever rely on the redundancy in any fashion.
Posted by uninet, 11-24-2016, 12:48 PM
That's very helpful, Tony. Thank you. I was wondering if I was making a mistake if I skipped RAID. I use CPanel's backup system to store the daily backups of clients on Amazon S3. I've been debating if real time redundancy with a RAID1 was really needed with SSDs, though. Good point on the cache drive. That's right. I'd looked into that for my home server (which is a RAID-Z), but didn't pursue it. Thanks, Tim
Posted by Lakjin, 11-24-2016, 08:49 PM
This is extremely terrible advice. Please, dont listen to this guy. The solution to people using RAID as a substitute for backup is not to stop using RAID, it is to educate clients. RAID and backup are not substitutes, and both have their uses (and shortcomings). For anyone that cares about uptime, RAID is absolutely necessary. Is it a perfect solution? No, but nothing is. However, it helps keep servers up when hard drives do fail. (Assuming we are talking about mirroring RAIDs.)
Posted by ayksolutions, 11-25-2016, 01:34 PM
You can also consider this as an option: https://romanrm.net/mhddfs.
Posted by uninet, 11-25-2016, 02:19 PM
So, to take my question in a slightly different direction: ServerHub is running an incredible Black Friday special on E3-1230v2's. Obviously, that's actually a downgrade from my current server, but they are charging only $215/quarter for one with 2x500GB hard drives, 32GB RAM, 1Gbps port, cPanel, and full management (which I wasn't looking for, but hey, if someone else can loose sleep over things...). That got me thinking: what if I bought two of those and somehow split the load on my server? Not all of my eggs would be in one basket that way. Is there a good way, for example, that I could put ASSP and MySQL on one server and have web and mail on the other? (Within the context of cPanel still functioning normally, that is -- I know I could without cPanel, but I do want to stick with a control panel.) Part of me likes the idea of having two slightly less powerful servers rather than 1 more powerful server. Thoughts?
Posted by ayksolutions, 11-25-2016, 03:30 PM
I have no experience with Serverhub but be sure to check for reviews on them, contact them to see how they reply, etc.. Good luck.
Posted by madRoosterTony, 11-25-2016, 04:36 PM
Well after 15+ years of experience with servers, I can honestly say more then 60% of the time in my experience with RAID over the years, when a hard drive fails it takes the entire raid cluster out with it and there was zero redundancy. There are many cases reported here on WHT, that one drive in a RAID system has failed and corrupted all the data across all the drives. Using RAID for redundancy is the stupidest thing I personally have ever seen and was not what RAID was originally designed for, it was just a side effect that everything went 100% right in a hard drive failure, you had redundancy. RAID's whole purpose in life is to either provide better performance or to provide a larger single disk space. I have seen this data corruption in all types of RAID (Linux Software, ZFS, Hardware, Hardware with battery backup) and have also seen clients loose their mind when they realize that redundancy was not what they got in their RAID system. RAID definetly has its place in servers, but anyone that says its redundancy, needs more experience and research with it.
Posted by madRoosterTony, 11-25-2016, 04:39 PM
Cluster cPanel is a complete nightmare. It can be and has been done, but be prepared to have hours spent getting everything to sync between both servers and then constantly having to tweak it as cPanel breaks it with updates. You would be better off splitting your accounts between the two servers creating an smaller load on each.
Posted by uninet, 11-25-2016, 04:56 PM
I had a feeling, but it *seemed* like a good idea. I have learned over the years that trying to do anything cPanel doesn't officially support is generally a pain, though. Hmm... splitting the load might still make sense. I think the load is relatively concentrated on a few accounts (including my own), but...
Posted by uninet, 11-25-2016, 07:03 PM
Hey, Steve, is the HiVelocity Sales Team off for the day? I tried e-mailing a few questions this afternoon and didn't hear back. Live chat seems to indicate no one is there. Just wondering since your special ends tonight...
Posted by hivelocitygm, 11-25-2016, 09:00 PM
We let the Sales Team cut out a little early tonight. We sold a ton of servers and our build team is close to overwhelmed. We will keep the deals up until Monday morning. If you purchase a server before Midnight Tonight we will drop the puck for you Monday (we will email you to schedule your Plinko drop). If you purchase this weekend you will get the good price but no bonus deal. Our Sales Team will be available via chat or phone tomorrow morning if you have any questions.
Posted by uninet, 11-25-2016, 09:53 PM
Rats, that's too bad (although kind of you for your employees). I had a question outstanding from a couple of days ago with the sales rep I was talking to. I figured he probably had Thanksgiving off and I didn't want to be a pest with more questions until I heard back from him, so I didn't write immediately this morning either, but tried to leave some time for you guys to get caught up. I waited until about 2:15 Eastern Time to follow up further with him, but apparently waited too long.
Posted by hivelocitygm, 11-26-2016, 09:28 AM
Forward me the question and I will make sure we get an answer to you this morning. Sorry about that.
Posted by uninet, 11-26-2016, 04:11 PM
No problem! I know things happen. Rob replied in place of Kevin this morning to all but one part. I'll send that to you in a moment. Thanks!
Posted by NortheBridge, 11-26-2016, 11:19 PM
That would explain why when some people from our suppliers department called all they got was fifteen minutes of waiting to voicemail. Although they called during the day Pacific time I guess it could have been late enough to be considered the late afternoon Eastern. Adding servers in Florida to extend a few services have been a side project but after the first several attempts to contact Hivelocity we stopped and moved on. It's great that you let your team leave a little early on Friday but when you get messages saying they are still there only to be routed in circles it doesn't leave a good impression. So the new vSphere clusters in Florida is going to have to wait until the next time we get around to it. Needless to say, we will still put forth recommendations for HiVelocity because you all seem like an honest bunch that provide good services.
Posted by Get-Acom, 11-26-2016, 11:25 PM
before you go to hive. I suggest you check out the association btwn, Hivelocity, Colocrossing, Hudson Valley Host, Datashack check the blacklist of them all. check each reputation here, and other places. Then re-weigh the offer.
Posted by NortheBridge, 11-26-2016, 11:41 PM
Rather than a "mysterious post" with not detail but a list of providers @Get-Acom, what exactly are you trying to insinuate?
Posted by hivelocitygm, 11-27-2016, 11:20 AM
The Sales Team was around until about 6pm Friday evening EST. If you called afterward and sat on hold I am super sorry. I know that can be extremely frustrating. If I can make it up to please email me directly and I will make sure we do our best to make giving us a shot worth your while. I don't normally let the sales team shut down early but with our Black Friday deals combined with a single customer ordering close to 200 servers our Server Deployment Team needed to come up for air. Once again, if you decide to give us another shot now or down the road, please loop me in with an email so I can be sure we put our best foot forward.
Posted by hivelocitygm, 11-27-2016, 11:24 AM
I cannot speak to the other companies you have listed but I can say unequivocally we have no association with any of the providers you have listed, nor do we have any blacklist issues. Our Abuse Dept runs a tight ship which you will find spammers complaining about if you do just a little research. I am curious, however, how you came to the conclusion that we are affiliated with ColoCrossing, Hudson Valley or DataShack.
Posted by hivelocitygm, 11-28-2016, 01:16 PM
I am very curious to understand how this theory was put together as well. I can assure you it is pure conjecture and no truth to it but I would love to understand the motivation.
Posted by NortheBridge, 11-29-2016, 01:25 AM
@hivelocitygm - according to our Verizon Enterprise Global IP/PSTN Hybrid Switch logs, the last call attempt was at 2:33pm PST so that would have been pretty close to 6pm EST lasting 5 minutes and 14 seconds. Hey, it was a Friday, a holiday weekend, we even let our people go early even if they weren't in the United States so I completely understand. Now we have all the information versus 'well, no one is picking up even though it says they are open and chatting is routing in circles' so that does clear the air. We will definitely give you all another shot down the road; Florida is one of the hardest markets to find reliable partners so we always end up bouncing traffic to Washington, D.C. which is a major hub for us but future operations are going to depend on having availability in Florida due to the nature of our primary business which is not hosting. I will definitely loop you in via email the next time around as our cluster deployments can be complicated and ultimately end up being described as "mini-DCs" when they are completed. We may never know but this is an example of why we do our own deep in-depth research. I used the term 'insinuation' but it was definitely an 'accusation' of something. It could possibly be something as simple as a disgruntled customer which happens.

Add to Favourites Print this Article